U N I V E R S I T Y O F C O P E N H A G E N F A C U L T Y O F S C I E N C E

The effects of riverine DOM on microbial composition and function

Elisabeth Münster Happel

Marine Biological Section

Department of Biology

Title: The effects of riverine DOM on microbial composition

and function Author: Elisabeth Münster Happel Submission: This thesis has been submitted June 30th 2018, to the PhD

School of the Faculty of Science, University of Copenhagen, Denmark

Supervisor: Professor Lasse Riemann University of Copenhagen, Denmark Co-supervisor: Associate Professor Veljo Kisand University of Tartu, Estonia Committee: Professor Chris Francis

Stanford University, USA

Academy Research Fellow (adjunct professor) Susanna Hietanen University of Helsinki, Finland

Professor Anders Priemé University of Copenhagen, Denmark

Frontpage photo from: https://pngtree.com/freepng/water_507742.html

The effects of riverine DOM on microbial

composition and function

Resumé

Mikroorganismer spiller en vigtig rolle i omsætningen af opløst organisk materiale i både marine og kystnære områder. I kraft af deres enorme diversitet og energiomsætning, driver de de biogeokemiske kredsløb, der er vitale for alle levende celler. Kvælstof er vigtig for alt liv og det er oftes kvælstof, der begrænser biomasseproduktionen i vandmiljøer. Det er en lang række forskellige mikrobielle grupper, der driver de processer, der tilsammen udgør kvælstofkredsløbet, men der er stadig meget, vi ikke ved om, hvordan disse processer bliver påvirket af menneskeskabte aktiviteter specielt i kystnære områder. Østersøen er et kystnært område, som er under stadig stærk påvirkning af menneskelige aktiviteter. Kvælstofkredsløbet er særligt vigtigt her, da det menes at være ansvarlig for at fjerne en betydelig del af det kvælstof, der kommer fra land, inden det møder det åbne hav. Fremtidige ændringer i nedbør menes at øge tilstrømningen af flodvand og derved også opløst organisk materiale til de kystnære områder i Østersøen. Derfor behandler denne afhandling effekten af opløst organisk materiale fra floder på det mikrobielle samfunds sammensætning og funktion. Den overordnede sammensætning af gener, samt sammensætningen af vigtige funktionelle gener relateret til kvælstofkredsløbet, er undersøgt gennem både eksperimentelle forsøg samt prøvetagninger jævnt fordelt ud over de meget forskelligtartede kystområder i Østersøen. Sammensætningen af funktionelle gener bliver her brugt som indikator for funktionelt potentiale i økosystemet. Resultaterne viser, at selvom de mikrobielle samfund er meget forskellige i de undersøgte områder af Østersøen, så responderer de alle på opløst organiske materiale fra floder, når man ser på sammensætningen af gener. Selvom der ikke var en overordnet effekt af flodvandet på den mikrobielle sammensætning samt sammensætningen af kvælstofrelaterede gener, så var der effekter relateret til specifikke taxonomiske grupper og processer. Endvidere kunne vi påvise potentielle funktionelle effekter af flodvandet i kraft af ændringer i genudtryk fra et vigtigt kvælstofrelateret gen.

Samlet set indikerer disse resultater, hvordan opløst organisk materiale fra floder kan forme mikrobielle samfund, samfundets funktionelle profil samt essentielle gener involveret i kvælstofkredsløbet. Disse resultater illustrerer, hvordan funtionelle gener potentielt kan være brugbare indikatorer for miljømæssige forandringer og derigennem i fremtidig miljømonitorering.

1

2

Abstract

Microorganisms play a crutial role in remineralizing dissolved organic matter (DOM) in both oceanic and coastal environments. Due to their vast diversity and metabolic capability, they drive the biogeochemical cycles that are vital to every living cell. Nitrogen (N) is a key constituent of cells and usually limits production in aquatic environments. Diverse microbial assemblages mediate the different steps of the N cycle. There are large gaps in our knowledge on how these key processes are affected by anthropogenic pressure, especially in coastal zones. The Baltic Sea is a large coastal system which is heavily affected by anthropogenic activities. N cycling is particularly important here as it is thought to remove a substantial part of N coming from land before entering the open ocean. Projected changes in precipitation is believed to increase river discharge and thus river DOM to the Baltic Sea coastal zones. The works of this thesis address the effects of riverine DOM on microbial composition and function. Over-all community gene composition and key functional genes involved with N cycling were investigated through microcosm experiments and environmental sampling covering the highly diverse coastal zones of the Baltic Sea. The composition of functional genes are, here, used as a proxy for potential ecosystem function. The results show that although communities from different areas of the Baltic Sea are distinct, there are functional responses in gene composition to riverine DOM. Although there was no over-all response to riverine DOM in microbial composition and the composition if N cycling genes, deducible responses were linked to specific taxa and processes. Further, expression of a key functional N cycling gene revealed possible active functional responses to different riverine DOM.

Collectively, the findings of this thesis provide indications of how riverine DOM can shape microbial communities, community functional profiles and key N cycling processes. These results illustrate the potential suitability of functional genes as predictors of environmental change and thus for future environmental monitoring.

3

4

Acknowledgements

Firstly, I would like to give my warmest thank you to my Phd supervisor Lasse Riemann who has been a great support during the years that I spent in his lab. I couldn’t have hoped for a better supervisor. I am very thankful for your patience and for everything you taught me. You are not only a great person but also a great role model and someone I look very much up to. I also have to give warm thanks to some of all the great people that I met during my time in the LaR lab. Sachia J. Traving, Deniz Bombar, Ryan Paerl, Sophie Charvet, Daniel Ayala, Mar Benavides and Søren Hallstrøm; I learned a lot from all of you and you have my greatest admiration. A special thanks to one of the nicest people I’ve ever met, Sachia – I really appreciate the numerous coffee-break discussions, your guidance on statistics and bioinformatics and all the fun times we’ve had both in and outside the lab. I would also like to thank my co-supervisor Veljo Kisand for constructive discussions, guidance and for helping me with all the bioinformatics. I also give my warmest thank you to all the Blueprint people, especially to the microcosm people: Trine Markussen, Mathias Middelboe, Jeanett Hansen, Veljo Kisand, Jonna Teikari and Vimala Huchaiah. Thanks for all the hours spent, setting up the experiments, sampling and filtering – you made it all a fun experience. Thank you to Anders Andersson and Johannes Alneberg for all your help with the BARM database. Thank you to Niels Daubjerg for letting us use your lab for molecular work and to Tvärminne station for accommodating us during the experiments. A special thanks to all the amazing people I met in Woods Hole during the Microbial Diversity course. I had a great time and I’m very grateful that I got the opportunity to meet so many nice and skilled people. I would like to give warm thanks Jonathan Zehr for welcoming me to visit his lab during my scientific exchange. Also a special thanks to Britt Henke and all the great people I meet in the Zehr lab – you made me feel very welcome. A warm thank you to Daniel Castillo Bermúdez, Mar Benavides, Sachia J. Traving and Beata Lehka for reading and commenting my thesis – I admire you all very much so it was a great honour that you offered to help me. A big thank you to all the people at MBS

Lastly, I have to thank my family for all your support. I look forward to spending much more time with you now.

This work resulted from the BONUS Blueprint project supported by BONUS (Art 185), funded jointly by the EU and the Danish Council for Independent Research, Estonian Research Council, Swedish Research Council FORMAS, and Academy of Finland.

5

6

List of included papers

I. T. Markussen, E. M. Happel, J. E. Teikari, V. Huchaiah, J. Alneberg, A.

Andersson, K. Sivonen, L. Riemann, M. Middelboe and V. Kisand (2018)

Coupling biogeochemical process rates and metagenomic blueprints of coastal

bacterial assemblages in the context of environmental change. Submitted to

Environmental Microbiology

II. E. M. Happel, T. Markussen, J. Teikari, V. Huchaiah, J. Alneberg, A. Andersson,

K. Sivonen, M. Middelboe, V. Kisand and L. Riemann (2018) Effects of

allocthonous DOM input on microbial composition and nitrogen cycling genes

at two contrasting estuarine sites. Submitted to Environmental Microbiology Reports

III. E. Happel, I. Bartl, M. Voss and L. Riemann (2018) Extensive nitrification and

active ammonia oxidizers in two contrasting coastal systems of the Baltic Sea.

Environmental Microbiology, in press. Doi: doi.org/10.1111/1462-2920.14293

7

8

Contents

Resumé 1

Abstract 3

Acknowledgements 5

List of included papers 7

Introduction 11

Aims of the thesis 21

Discussion 23

Future perspectives 29

References 31

Paper I 41

Supporting Information – Paper I 67

Paper II 81

Supporting Information – Paper II 95

Paper III 101

Supporting Information – Paper II 123

9

10

Introduction

Microbial communities and ecosystem functions

Microorganisms (here referring to bacteria and archaea) are the most abundant and

diverse component in aquatic environments. Abundances may reach 105 cells ml-1 in

aquatic environments and an estimated total of 1029 in the global oceans (Whitman et al.,

1998). The biogeochemical cycling of essential elements like carbon (C) and nitrogen (N)

is carried out by microorganisms, which harbour a diverse range of metabolic processes

that drive the cycling of elements (Galloway, 1998; Falkowski et al., 2008). The majority

of C present in the oceans is in the form of dissolved organic matter (DOM) which is a

chemically complex mixture of amino acids, lipids, sugars, humic and fulvic acids

(Benner, 2002). Microorganisms play a key role in remineralizing DOM. Microbial

processing of DOM is also coined the ‘microbial loop’ (Fig. 1) (Azam et al., 1983; Azam,

1998) and is an ecological important process which recycle DOM and make nutrients and

energy available to higher organisms. DOM mineralized in the ‘microbial loop’ fuel

regenerated production in nutrient limited waters (Fenchel, 2008). As microbial

communities process DOM, they also affect the composition of DOM (Kujawinski et al.,

2016). Marine DOM is estimated to contain ~700 x 1015 g carbon, which is the equivalent

of the atmospheric carbon pool stored as CO2 (Hedges et al., 2002); and understanding

the mechanisms regulating the DOM pool is crucial for our understanding of aquatic

ecosystem function and the global C cycle.

11

Figure 1. The ‘microbial loop’ illustrating the cycling of matter between different trophic levels (Azam et

al., 1983).

The role of microbial diversity might yield insights into the potential functions and

elasticity of ecosystems. For decades, the 16S rRNA gene has been used as a phylogenetic

marker for investigating microbial communities which has elucidated a remarkable

diversity of organisms (Woese and Fox, 1977; Woese, 1987). Technological advances have

provided low-cost next generation sequencing (NGS) which have allowed for deeper

sequencing of microbial communities (Kircher and Kelso, 2010). NGS data has

demonstrated that not only is microbial diversity great, but that there is a long tail of rare

taxa that could function as a ‘seed bank’ of metabolic potential (Pedros-Alio, 2006; Sogin

et al., 2006). There is great variation in microbial diversity across environments (Walsh et

al., 2015). Further, the evenness of communities has been shown to have an impact on

functional stability (Wittebolle et al., 2009). Despite the increasing information

accumulated through NGS data, big gaps remain in our knowledge of how microbial

community composition and diversity affects ecosystem function (Fuhrman, 2009;

Bernhard and Kelly, 2016). To bridge the gap between ecosystem functions and

taxonomy, a great number of recent studies have turned their focus onto a trait based

approach (Green et al., 2008) and the use of functional gene as marker for specific

processes (reviewed in Imhoff, 2016). The use of functional genes provides a functional

centric approach on investigating microbial diversity. Moreover, the highly targeted

approach of using functional genes improves the resolution of the community (especially

12

of low abundance taxa). The faster rates of evolution (compared to 16S rRNA) also

provides a higher phylogenetic resolution making it easier to discriminate closely related

groups (Imhoff, 2016). Although marine microbial diversity is great, only few conserved

core functional genes have been proposed to code for the redox reactions that drive

biogeochemical cycles (Falkowski et al., 2008). Metagenomics provide large amounts of

information allowing whole-community investigations of functional gene profiles, which

provide holistic insights into, not only taxonomic composition, but also the functional

potential of ecosystems. The un-targeted (PCR free) approach of shot-gun sequencing has

greatly contributed to new discoveries of key players in N cycling i.e. the potential

connection of the widely distributed Planctomycetes to N2 fixation (Delmont et al., 2018).

Another discovery has been archaeal ammonia monooxygenase-like genes in

metagenomes from the Sargasso Sea (Venter et al., 2004), a function which previously has

been assumed to be exclusively located within bacteria. The presence of archaeal

ammonia oxidizers (AOA) was later confirmed (Könneke et al., 2005) and now we know

that AOA are widespread in many environments (Leininger et al., 2006; Veuger et al.,

2013; Beam et al., 2014; Erwin et al., 2014).

N cycling – Players and processes

N is as well as C, a key constituent of every living cell. It is the building block of e.g.

nucleic acids and proteins that drive all metabolic processes in cells (Galloway and

Cowling, 2002). The largest pool of N on earth exist as dinitrogen gas (N2) which makes

up ~80% of the atmosphere (Galloway and Cowling, 2002) and the dominant form of N

dissolved in the oceans (Karl et al., 2002). N2 is inert and can only be utilized by nitrogen

fixing organisms (diazotrophs) that are able to fix the N2 into ammonia (NH3) thus

making the great pool of N available to other organisms (Postgate, 1970). In fact, a great

deal of earths biodiversity has evolved through competition under N-limited conditions

(Galloway and Cowling, 2002).

13

Despite the great amount of N in the oceans, the available forms of N are often limiting

organic productivity (Ryther and Dunstan, 1971; Falkowski, 1997; Karl et al., 2002;

Rabalais, 2002; Glibert et al., 2016). Therefore, the effects of the availability of N on system

productivity has been intensely studied (Haines and Wheeler, 1978; Duce, 1986; Wheeler

and Kirchman, 1986; Bradley et al., 2010).

Figure 2. Simplified representation of the major N cycling processes and the different oxidation states of

N. Grey arrows indicate anaerobic processes (Gruber, 2008).

N is present in the environment in many different forms and oxidative states, from the

most reduced form (NH4+) to the most oxidized form (NO3-) which means that it can be

used as both electron donor and acceptor in energy metabolism (Fig. 2) (Gruber, 2008;

Stein and Klotz, 2016). The number of different transformation processes makes the N

cycle one of the most complex cycles; a cycle that also influences the cycling of both C

and phosphorous (P). Most N transformation processes are mediated by bacteria or

archaea as part of their metabolism; either through energy acquisition or for synthesis of

structural components (Richardson and Watmough, 1999; Gruber, 2008; Dang and Chen,

2017). Major processes include biological N2 fixation (Wilson, 1957), nitrification

(Schloesing and Muntz, 1877) and denitrification (Kluyver and Donker, 1926).

14

As mentioned above, N2 fixation fixes N2 gas into NH3 by the use of the enzyme

nitrogenase (Boyd and Peters, 2013). The nitrogenase enzyme complex is inhibited by O2

(Goldberg et al., 1987) which means that phototrophic diazotrophs (cyanobacteria) have

evolved strategies of either temporal or spatial separation (by heterocyst formation;

Haselkorn, 1978) of N2 fixation and the O2 producing photosynthesis. A diverse group

of bacteria and archaea are able to fix N and the marker gene nifH (coding for the iron-

subunit of the nitrogenase enzyme complex) has been amplified from most aquatic

environments (Zehr and McReynolds, 1989; Kirshtein et al., 1991; Ben-Porath and Zehr,

1994; Zehr et al., 2003; Farnelid et al., 2011). Abiotic N fixation also occurs through

atmospheric deposition creating nitric oxide (NO) and nitrous oxide (NO2) forming nitric

acid (HNO3) that falls as rain (Singh and Agrawal, 2008).

NH4+, NO3- and nitrite (NO2-) can be assimilated into organic N compounds (biomass)

by both phytoplankton and bacteria (Wheeler and Kirchman, 1986; Glibert et al., 2016).

While NH4+ can be taken up directly, NO2- and NO3- has to be reduced using either

nitrite or nitrate reductases (Richardson and Watmough, 1999). NO3- reduction to NO2-

is done by bacteria using three distinct enzymes depending on the purpose; for

assimilation (the cytoplasmic, Nas), for respiration (using nitrate as electron acceptor,

Nar) and for dissimilatory reduction (periplasmic Nap) (Moreno-Vivián et al., 1999).

Nitrite reductases (Nir) catalyse reductions of nitrite to either NO or NH4+ for either

respiration, assimilation or detoxification (Richardson and Watmough, 1999). Many

microorganisms are also able to utilize dissolved organic N (DON) (Seitzinger and

Sanders, 1997) and urea; through urea hydrolysis (Solomon et al., 2010).

NH3 can be oxidized by groups of both ammonia oxidizing bacteria (AOB) and

archaea (AOA) via hydroxylamine (HAO) into NO2- and further to NO3- by nitrite

oxidizing bacteria (NOB). This process, collectively named nitrification with the first step,

ammonia oxidation, considered the rate-limiting step. Ammonia oxidation is carried out

by use of the ammonia monooxygenase (AMO) enzyme. The marker gene for ammonia

oxidation is the structural ammonia monooxygenase gene, amoA, which has been used to

characterize and quantify both AOB (Rotthauwe et al., 1997) and AOA (Francis et al., 2005;

Könneke et al., 2005; Treusch et al., 2005) in aquatic environments. Recent discoveries have

15

revealed organisms able to perform the complete oxidation of NH4+ to NO3-, a process

named comammox (Daims et al., 2015; van Kessel et al., 2015).

In the absence of oxygen, oxidized N compounds can be used as electron acceptors by

bacteria. Denitrification is the step-wise reduction of NO3- to NO2- or N2, thereby

completing the N cycle (Fig. 2). The different reduction steps are carried out by diverse

groups of denitrifiers including both bacteria, archaea and fungi (Zumft, 1997; Philippot,

2002; Heylen et al., 2006; Wei et al., 2015). The key enzymes in denitrification are the

dissimilatory nitrate reductases (Nar and Nap genes), two types of nitrite reductases (nirK

and nirS) and nitric oxide reductase (norZ) (Heylen et al., 2006). Other anaerobic processes

include anaerobic ammonia oxidation (anammox) and dissimilatory nitrate reduction to

ammonia (DNRA). Anammox refers to the coupled oxidation of NH3 and reduction of

NO2- (Mulder et al., 1995). Anammox is carried out by both bacteria and archaea (Francis

et al., 2007). In the open oceans, Anammox is believed to contribute to 50% of the N2

produced (Jetten et al., 2009). DNRA, which is performed by both bacteria and fungi

(Stein and Klotz, 2016) also reduces NO3- but rather than contributing to N loss it

conserved N in the system and in coastal zones, it is believed to contribute to ~30% of

NO3- reduction.

Although relationships between environmental factors and N cycling communities

and functions have been studied in various environments there are still large gaps in our

knowledge on what constrains key processes (HELCOM, 2009; Damashek and Francis,

2018; Kuypers et al., 2018). Collectively, N cycling processes regulate the availability of N

in marine ecosystems. Global budgets on N inputs and N loss are out of balance, meaning

that estimated global N loss exceeds that of N inputs (Galloway et al., 2004). Explanations

for this imbalance are still unclear but heavily debated (Falkowski, 1997; Tyrrell, 1999;

Codispoti et al., 2000; Jickells et al., 2017). The introduction of reactive N through

anthropogenic activities now dominate N budgets in all systems (Galloway et al., 2004;

Gruber and Galloway, 2008). The input of N to global oceans is at an estimated 87-156

Tg N year-1 from biological N2 fixation, 48 Tg N year-1 from rivers and 33 Tg N year-1

through atmospheric deposition. N losses are estimated to be around 150-450 Tg N year-

1 through denitrification, 4 Tg N year-1 through N2O production and 14 Tg N year-1

16

through deep sediment deposition. The input of N from rivers is increasing over time,

greatly influencing coastal ecosystems. However, most of the N is converted to N2 before

entering the open oceans (Galloway et al., 2004). Ultimately anthropogenic activities,

through introduction of more reactive N into environments primarily through

agriculture but also indirectly due to climatic changes, means that the global N cycle is

undergoing great changes (Galloway and Cowling, 2002; Rockström et al., 2009).

Anthropogenic effects on coastal zones

Coastal waters are among the most valuable ecosystems on the planet with regard to e.g.

fish and shellfish production, recreation, and waste assimilation. It is therefore of

fundamental importance to understand and predict the productivity and stability of

these ecosystems.

One of the greatest influences humans has had on N cycling has been the industrial

fixation of N and subsequent use of N fertilizers in agriculture (Galloway et al., 2004). In

1908 the Haber-Borsch process of synthesising NH3 was developed (Smil, 2001; Erisman

et al., 2008). This process is now responsible for feeding ~48% of the human population

(Stein and Klotz, 2016). Since 1965, the amount of industrial N2 fixation resulting in new

reactive N has exceeded that of natural terrestrial production (Galloway and Cowling,

2002). Not only has the amount of N fertilizers increased but the composition of fertilizers

has also changed. Now urea is used in >50% of fertilizers globally, surpassing NO3-

(Glibert et al., 2006).

The intensified applications of N rich fertilizers and climatic changes predicted to

cause increased precipitation (Trenberth, 2011) means that the flux of both dissolved

inorganic N and DOM from land to coastal zones likely increase. This will affect both

surface run-off (diffuse sources) and discharges from rivers (point sources) to coastal

zones. DOM input from rivers serves as an important source of labile N in estuaries

(Seitzinger et al., 2005; Bronk et al., 2007; Knudsen-Leerbeck et al., 2017). It is estimated

that global inputs of N through rivers has increased from 27 Tg N year-1 in 1860 to 47.8

17

Tg N year-1 (Galloway et al., 2004). The change in both N composition and amount will

likely have a great impact on coastal recipient communities and especially on N cycling.

Further, increases in inorganic N (DIN) has been shown to affect both DOM composition

(lability) and microbial community composition (Goldberg et al., 2017). Overall increases

in reactive N in coastal zones has great impacts on the composition, productivity and

functions of these systems (Vitousek et al., 2002).

The Baltic Sea – A case on anthropogenic pressures on coastal zones

There is an immense interest in limiting anthropogenic perturbations and especially

eutrophication in the Baltic Sea (Bonsdorff et al., 1997) as it has potential consequences

including algal blooms of possibly harmful algae (HAB), anoxia and reductions in fish

stocks which are detrimental to the future economic prosperity of the entire region

(HELCOM, 2009)

The Baltic Sea is the world’s largest semi-enclosed brackish water system (Bernes,

2005). It consists of several sub-basins with unique geology (Rönnberg and Bonsdorff,

2004). The Baltic Sea receives freshwater from nine countries and an additional six

countries in upstream catchments (Arheimer et al., 2012). There are more than 200 rivers

feeding into the Baltic Sea including nine major rivers. There are great differences in these

river catchments ranging from agricultural in the south to forest and peatland in the

north. The sub-basins of the Baltic Sea are influenced by different river catchments with

varying DOM, humic substances and metal concentrations depending on catchment

characteristics, drainage area and discharge rates (Pettersson et al., 1997). The freshwater

input from rivers and the connection to the North Sea via Kattegat has created a strong

salinity gradient from ~20 to ~3 (Kullenberg and Jacobsen, 1981). The microbial

community of the Baltic Sea has been shown to follow similar gradient changes

(Herlemann et al., 2011; Herlemann et al., 2016). In addition, the microbial communities

of the Baltic Sea have been shown to be similar to other brackish water communities

(Hugerth et al., 2015). Most of the southern Baltic Sea and the Baltic Sea proper has low

18

DOM concentrations and is considered N limited while the northern part has high DOM

concentrations and is P limited (Bernes, 2005; Rowe et al., 2018). Low N:P ratios is known

to favour nitrogen fixation in the Baltic Sea (Niemi, 1979). During the summer, the Baltic

Sea is subjected to re-occurring cyanobacterial blooms, a characteristic feature of the area

(Stal et al., 1999; Bernes, 2005). Microbial communities of Baltic Sea sediments (Vetterli et

al., 2015) and surface water follow seasonal successions connected with phytoplankton-

blooms and the inflow of freshwater populations from rivers (Laas et al., 2015).

The deep basins of the Baltic Sea are characterized by a stable chemocline separating

the oxic surface water from the suboxic bottom water. These chemoclines are

characterized by strong stratification of taxa as well as functional capacities. The

communities are enriched in genetic potentials for copiotrophic lifestyles reflecting

possible adaptations to eutrophication (Thureborn et al., 2013). Apart from seasonal

cyanobacterial blooms, heterotrophic N2 fixation has also been detected at and below the

chemocline (Farnelid et al., 2013b). The chemoclines are sites of extensive N cycling and

nitrification is particularly high in the upper layer of the chemocline (Jäntti et al., 2018)

where AOA play an essential role (Thureborn et al., 2013; Berg et al., 2015). Nitrification

also occurs in both sediments (Hietanen and Lukkari, 2007) and throughout the water

column (Bartl et al., 2018) of the coastal Baltic Sea. The Baltic Sea coastal zone functions

as an important filter, which through coupled nitrification-denitrification, can remove a

large fraction of the N entering the area via rivers (Silvennoinen et al., 2007; Voss et al.,

2011). Denitrification occurs both in the water column and in the sediments (Gran and

Pitkänen, 1999) but with much lower but stable rates in shallow sediments (Hietanen et

al., 2012). Both denitrification, DNRA and anammox occurs in the chemocline, with

highest rates from denitrification (Hietanen et al., 2012; Bonaglia et al., 2016). Anammox

also occurs in the sediments where it can be responsible for as much as 10-15% of N2

production (Hietanen and Lukkari, 2007). In sediments, a substantial part of the NO3-

removal is believed to go through DNRA (Karlson et al., 2005). Denitrification is

estimated to remove as much NO3- as the amount of NO3- entering the Baltic Sea through

rivers (Voss et al., 2011). Terrestrial organic matter makes up a substantial part of the

19

DOM pool in the Baltic Sea and >50% of the high-molecular weight DOM from rivers is

thought to be removed in the coastal zone (Deutsch et al., 2012).

There has been a fourfold increase in total N and eightfold increase in P to the Baltic

Sea due to anthropogenic activities in the last decades causing eutrophication and anoxia

(Larsson et al., 1985; HELCOM, 2009). Coastal N sources include inputs from rivers,

terrestrial run-off, atmospheric deposition and remineralisation from sediments (Bradley

et al., 2010). Of the N that enters the Baltic Sea via rivers, 48% is as DIN, 41% DON and

11% as particulate organic N (PON). The bioavailability of riverine DON is high (~30%)

and varies greatly between rivers (Stepanauskas et al., 2002). It is estimated that 52% of

the N originates from agriculture catchments and likewise 31% of the N comes from

Poland, a country with extensive agricultural production (Arheimer et al., 2012).

Future projections for the Baltic region suggest up to 30% increase in precipitation and

as a consequence of this, an increased discharge riverine DOM into the Baltic Sea, especially

in the northern parts (Andersson et al., 2015). Due to the different catchments in the Baltic

Sea, the riverine DOM is likely to differ between the forested north and the agricultural

south. How these increased inputs of river water and thus, riverine DOM, will affect the

microbial composition and function, particularly in processes related to N cycling in the

Baltic Sea, is the focus of this thesis.

20

Aims of the thesis

The works of this thesis seek to contribute to the gaps in present knowledge on how

microbial diversity and community composition is linked to ecosystem functions. To

bridge these gaps, functional genes play an essential role together with coupling

functional gene expression and ecosystem process rates. In light of climatic changes and

the projected increases in riverine discharge into the Baltic Sea, this was chosen as our

study-site. The hydro-geographical heterogeneity of the Baltic Sea is ideal for

investigating the effects of the riverine DOM of different catchments on microbial

community composition and function. The composition of functional genes are used as

a proxy for community metabolic potential and herein potential ecosystem function.

Shortly summarized, to assess the functional responses of microbial assemblages to

different sources of DOM, we carried out two controlled microcosm experiments where we

exposed natural community assemblages from distinct coastal sites in the Baltic Sea

Research questions of the thesis • If the characteristics of DOM shape microbial communities, can it also

shape the whole community functional profiles? (Papers I and II)

• Does different riverine DOM affect microbial N cycling? (Paper II)

• Are N cycling communities shaped by river catchment characteristics

and does this have an effect on their activity? (Papers II and paper III)

Collectively this will shed light on how increases in riverine DOM (through

increased precipitation) might affect coastal microbial communities and the

cycling of N in the Baltic Sea.

21

(Øresund and Storfjärden, Fig. 3) to DOM (from rivers of different catchments; Lapväärtti

and Lielupe River). We compared the functional gene compositional responses through

metagenomic sequencing. Metagenomic sequencing allowed an in-depth investigation of

functional gene responses involved with overall DOM utilization genes and genes involved

with N cycling. In order to investigate relationships between undisturbed natural microbial

communities and environmental characteristic connected to river catchments, we sampled

to distinct estuaries in the Baltic Sea (Bay of Gdańsk and Öre Estuary, Fig. 3). We examined

both the composition and activity of ammonia oxidizers and nitrification rates.

Figure 3. Map of the Baltic Sea; land use and sampling stations. Adapted from (Andersen et al., 2016).

22

Discussion

The effect of riverine DOM on microbial functional profiles

DOM has previously been connected with shifts in microbial community composition

and the abundance of specific taxa (Kisand and Wikner, 2003; Landa et al., 2015; Lindh et

al., 2015; Sipler et al., 2017; Traving et al., 2017). However, elucidating functional responses

from microbial processes involved in DOM utilization and the cycling of DOM

derivatives entails significant challenges. These challenges involve phylogenetic and

metabolic temporal and spatial variability, diversity and connectivity (McCarren et al.,

2010).

Generally, we found limited responses in the microbial community composition

(based on 16S rRNA genes) to the addition of riverine DOM (papers I and II). However,

when looking at and beyond the phyla-level, at a finer resolution (e.g. family level),

compositional differences were deduced, suggesting that the responses to riverine DOM

were indeed connected to specific taxa. Shifts in bacterial communities of the Baltic Sea

have previously been connected with river DOM inputs, however without any significant

differences between different river types (Traving et al., 2017). Similarly, the initial

communities of Øresund and Storfjärden were distinct, reflecting the heterogeneity of

Baltic Sea microbial communities (Herlemann et al., 2011; Herlemann et al., 2016; Hu et

al., 2016). The same distinct community structures (based on amoA genes) were evident

from the Öre Estuary and Bay of Gdansk (paper III). The lack of over-all phylogenetic

response, could suggest that the stability of these communities is driven by the

dominance of ‘generalists’ (Sriswasdi et al., 2017). The importance of generalists in

shaping microbial communities in coastal zones and in the Baltic Sea has previously been

established (Mou et al., 2008; Lindh et al., 2016).

To develop a mechanistic understanding of ecosystem functions herein the interplay

between microbial communities and physiochemical processes there is a need to look

beyond pure taxonomic composition and include metabolic functions (Krause et al.,

23

2014). We investigated the response in functional gene abundances through

metagenomic sequencing. Through metagenomics we demonstrate how whole

community functional gene composition or the ‘genetic blueprint’ can be linked to

differences in DOM sources (paper I). These findings provide great potential for the use

of whole community functional genes as predictors of environmental changes. The link

in responses of functional gene abundances to DOM types (paper I) was dominated in

function by the degradation and cycling of DOC and uptake systems (transporters) which

reflect interactions between cells and their surroundings. Transporters has been shown

to give insights into the ecological role of DOC turnover and their connection to specific

taxa (Poretsky et al., 2010). When looking at N cycling genes (paper II) there was no

universal compositional response to the different DOM types. A lack of response in the

composition of N related functional genes coupled with changes in community

composition could point to functional redundancy. However, there were responses in

functions related to specific processes (discussed below). Functional redundancy refers

to ‘species’ performing similar function in an ecosystem (Lawton and Brown, 1994) which

relies on the concept of ecological guilds (Root, 1967). Functional redundancy can enable

bacterial communities to maintain the ability to utilize DOM despite changes in

composition or reduced diversity (Sjöstedt et al., 2013). Recent studies have suggested

that environmental factors can be coupled to metabolic functions rather than taxonomy

and that microbial communities from various environments exhibit high functional

redundancy (Burke et al., 2011; Louca et al., 2016b; Louca et al., 2016a; Louca et al., 2018).

This could mean that a loss of diversity has little meaning for ecosystem function (Finlay

et al., 1997).

The use of whole-community functional gene profile might be a useful tool in

investigating responses of the microbial community, which may have great implications

for both element cycling, productivity and food web structures. A previous study from

the northern Baltic Sea found that addition of only inorganic N and P promoted

production of both bacteria, phytoplankton and zooplankton turning the system net

autotrophic, whereas the addition of humic carbon only affected the production of

24

bacteria and zooplankton, turning the system net heterotrophic. This could counteract

eutrophication in coastal waters (Andersson et al., 2013).

Riverine DOM on N cycling communities and processes

The use of functional genes as a means to examine diversity of specific ecotypes and their

correlations to environmental parameters has been used in a variety of environments

(reviewed in Imhoff, 2016). Most studies have thus far focused on single functional genes

such as nifH (N2 fixation)(Zehr and McReynolds, 1989; Kirshtein et al., 1991; Ben-Porath

and Zehr, 1994) or amoA (nitrification) (Rotthauwe et al., 1997; Horz et al., 2000; Francis

et al., 2005). These studies have been useful in providing information on diversity and

dispersal of such communities. However, only few studies have investigated several N

cycling genes or gene families collectively through metagenomics (Lu et al., 2012; Bristow

et al., 2015; Peura et al., 2015). Although we observed changes in whole community

functional responses (paper I), we only found minor changes in N cycling genes (paper

II). Instead of a universal N-related functional gene response to the different DOM types,

the composition of N related genes was distinct between the two sampling sites thus

highlighting the strength of local adaptations. Since the Baltic Sea is a system that has

been undergoing changes in N regime for decades, we are examining a system that has

already been under selective pressure for some time. Hence, the highly adapted N cycling

communities might not be able to respond to sudden changes in complex pools of DOM.

Further, the two distinct communities responded differently to the DOM types (in N gene

composition) which may not be too surprising given the different limitations of the two

sites (P limitation in the north, N limitations in the south).

We did observe responses to different DOM types of N cycling genes involved with

processes like ammonia uptake, N2 fixation and denitrification, showing that responses

in N cycling could be linked to specific processes. N2 fixation genes were over-

represented in response to the humic river water which could suggest favourable

conditions for heterotrophic N2 fixers (through increased DOC and changes in C:N

25

ratios). Heterotrophic N2 fixers are known from the Baltic Sea (Farnelid et al., 2013b;

Farnelid et al., 2013a) and their activity has been linked to organic C availability (Severin

et al., 2015). Some of the denitrification genes were also over-represented in response to

the humic river water. Studies have shown that expression of e.g nosZ is controlled by

copper availability (Sullivan et al., 2013). Humic river water likely carries metals like

copper due to their binding to humic substances (Benedetti et al., 1995).

The presence of functional genes or gene families merely indicate functional and

metabolic potentials. In order to examine actual functional activity, we used gene

transcripts as a quantitative indicator of a given process. Establishing links between gene

expression and process rates stem from the realization that genes code for proteins that

catalyse chemical reactions (Crick, 1958). However, for many N cycling processes,

correlations between gene and/or transcript abundances and the corresponding rates are

rare. Generally, significant correlations are more established for gene abundances than

for transcript abundances (Rocca et al., 2015). Also, there seems to be differences in the

strength of correlations depending on environment (Rocca et al., 2015). In soils, AOB

amoA gene abundances were the best predictors of nitrification rates, whereas nosZ and

nirK/S has been coupled to denitrification rates (Petersen et al., 2012). Relating transcript

abundances directly to process rates requires the assumption that the rate of translation

is the same as transcription. For bacteria, mRNA generally has a relatively short existence

due to post-transcriptional control enabling rapid adaptations to changing surroundings

(Richards et al., 2008). We focused on the expression of a core functional gene (amoA)

involved in nitrification. We quantified amoA genes and transcripts from both AOA and

AOB (papers II and III). Previously, a linear correlation was established between AOA

amoA gene abundances and nitrification rates (Smith et al., 2014) however, the

relationship between amoA transcript abundances and rates are still elusive. Despite this,

we found that even though several AOA groups are present in the Baltic Sea coastal

zones, the active fraction of the communities were dominated by a few groups across

sites (paper III). This highlights the need for examining not only the existing

communities but in particular the active fractions of such communities. Further, we

found discrepancies between gene and transcript abundances of AOA and AOB.

26

Although AOA was more abundant than AOB, AOB transcripts were most abundant.

Taken together, there is a need for additional studies on the mechanisms driving

ammonia oxidizing communities, amoA transcription and the relatedness to nitrification

rates.

We found a decrease in amoA transcript abundances in response to humic river water

DOM (paper II) suggesting that this type of DOM could have an inhibitory effect on

nitrification. Decreased rates of nitrification in coastal zones would greatly influence the

N removal and potentially lead to a higher flux of N from the coastal zones to open

oceans.

To examine whether N cycling communities could be shaped by local adaptations to

catchment types, we investigated the ammonia oxidizing communities of two distinct

river plumes (paper III). We used amoA as phylogenetic marker for ammonia oxidizing

communities. The usefulness of amoA relies on congruence between amoA- and 16 rRNA

phylogenies (Prosser and Nicol, 2008). Similar applications are frequently used in studies

of N2 fixers (based on nifH) (Hennecke et al., 1985). In studies of denitrifiers, representing

very diverse groups, not all functional genes are suitable phylogenetic markers, possibly

due to horizontal gene transfer (Salles et al., 2012). We found great differences between

the ammonia oxidizing communities at Bay of Gdansk and Öre Estuary suggesting that

the communities here are shaped by the local environments and thus perhaps the

catchment characteristics. This also affected the activity of the ammonia oxidizers

reflected in the higher nitrification rates at the Bay of Gdansk. Only few environmental

variables were correlated with both the composition of ammonia oxidizers, nitrification

rates (paper III) and functional gene composition (paper II).

There are still large gaps in our knowledge of how N cycling genes and communities

are constrained by environmental condition. The papers presented in this thesis do

however provide indications of how riverine DOM can shape, not only microbial

communities, but also the whole-community functional profiles (paper I), to some extent

the N cycling functional profiles (paper II) and the community composition and activity

of ammonia oxidizers (paper III). These responses reflect short-term perturbations, and

whether or not they could result in persistent changes remain to be tested. Some

27

predictive models do indicate that long-term regulation of the N cycle could be robust

and could quickly return to equilibrium after perturbations (Tyrrell, 1999).

28

Future perspectives

The results presented and discussed in this thesis are centred on the Baltic Sea, a highly

heterogeneous coastal environment. Indeed some of the findings highlighted how

microbial communities and their functional responses vary greatly within the Baltic Sea

coastal zone. Due to the long history of severe anthropogenic pressure on the Baltic Sea

and to the accumulated data collected by the scientific community for decades, the Baltic

Sea has been presented as a model for studying the future development of coastal oceans

(Reusch et al., 2018). Although the effects of increasing anthropogenic pressure to coastal

zones likely differ between systems, there are likely common trends that can be used to

aid a mechanistic understanding of how these areas will respond in the future.

The use of metagenomes to aid future environmental monitoring (Kisand et al., 2012)

may be particularly useful in the Baltic Sea. In the works of this thesis (papers I and II),

the BARM database (Alneberg et al., in press) was used as a reference to which

metagenomes could be mapped. This approach provides a fast tool for analysing large

and complex datasets as it eliminates the need for metagenome assembly. However,

when investigating functional genes like nifH and amoA, the usefulness of this approach

relies on the coverage of the database; how well represented these genes are in the

metagenomic backbone of the database. In addition, the use of protein families (PFAM)

and clusters of orthologs groups (COG) does present challenges. The protein families are

often categorized based on the structure or the function of the proteins and this can make

such categories unspecific when targeting specific functional genes. In example, COGs

related to nifH contain proteins that are unrelated to N2 fixation. Categorizing by genes

rather than protein families could solve this, however, there is a need for sufficient

functional gene databases.

To examine functional responses rather than mere compositional responses;

metatranscriptomics represents a promising tool for investigating over-all community

functional profiles of the active fraction of the community. In example,

Metatranscriptomics have been used to highlight the importance of particles in N cycling

29

processes (Ganesh et al., 2015). There are, however, still challenges to this approach.

Firstly, the transcriptomic profiles of marine microbial communities have been shown to

exhibit extensive diurnal oscillations (Ottesen et al., 2014). This challenge both sampling

(temporal) and connecting transcriptomic profiles to environmental parameters.

Secondly, although some data does exist on the presence of linear relationships between

functional gene and process rates (amoA abundances and nitrification rates (Smith et al.,

2014), researchers often struggle to connect transcript abundances and process rates in

environmental studies (Rocca et al., 2015). Efforts in combining laboratory studies of

isolates and natural assemblages with environmental studies is on-going, and rightfully

so. However, there is a need for improving the spatio-temporal resolution. Temporal

changes and oscillation in transcript abundances provide a number of challenges e.g.

how long mRNA persists in the environment, what regulates mRNA transcription and

translation (Moran et al., 2013).

The realization that the N cycle is much more complex than previously recognized

and the emergence of new knowledge on transformation processes and players in the

marine N cycle continues to push us closer to understanding how anthropogenic

activities affect ecosystems (discussed in Stein and Klotz, 2016 and Kuypers et al., 2018).

Since N limits primary production in the oceans it also limits the oceans ability to

sequester CO2 (McElroy, 1983; Falkowski, 1997) highlighting the need for continued

intensive research on N cycling organisms and processes.

30

References

Alneberg, J., Sundh, J., Bennke, C., Beier, S., Lundin, D., Hugerth, L. et al. (in press) BARM and BalticMicrobeDB, a reference metagenome and interface to meta-omic data for the Baltic Sea. Scientific data, in press.

Andersen, H.E., Blicher-Mathiesen, G., Thodsen, H., Andersen, P.M., Larsen, S.E., Stålnacke, P. et al. (2016) Identifying Hot Spots of Agricultural Nitrogen Loss Within the Baltic Sea Drainage Basin. Water, Air, & Soil Pollution 227: 38.

Andersson, A., Jurgensone, I., Rowe, O.F., Simonelli, P., Bignert, A., Lundberg, E., and Karlsson, J. (2013) Can Humic Water Discharge Counteract Eutrophication in Coastal Waters? PLOS ONE 8: e61293.

Andersson, A., Meier, H.E.M., Ripszam, M., Rowe, O., Wikner, J., Haglund, P. et al. (2015) Projected future climate change and Baltic Sea ecosystem management. Ambio 44: S345-S356.

Arheimer, B., Dahne, J., and Donnelly, C. (2012) Climate change impact on riverine nutrient load and land-based remedial measures of the Baltic sea action plan. Ambio 41: 600-612.

Azam, F. (1998) Microbial control of oceanic carbon flux: The plot thickens. Science 280: 694-696.

Azam, F., Fenchel, T., Field, J.G., Gray, J.S., Meyer-Reil, L.A., and Thingstad, F. (1983) The ecological role of water-column microbes in the sea. Marine Ecology Progress Series 10: 257-263.

Bartl, I., Liskow, I., Schulz, K., Umlauf, L., and Voss, M. (2018) River plume and bottom boundary layer – Hotspots for nitrification in a coastal bay? Estuarine, Coastal and Shelf Science 208: 70-82.

Beam, J.P., Jay, Z.J., Kozubal, M.A., and Inskeep, W.P. (2014) Niche specialization of novel Thaumarchaeota to oxic and hypoxic acidic geothermal springs of Yellowstone National Park. ISME J 8: 938-951.

Ben-Porath, J., and Zehr, J.P. (1994) Detection and Characterization of Cyanobacterial

Nifh Genes. Applied and Environmental Microbiology 60: 880-887.

Benedetti, M.F., Milne, C.J., Kinniburgh, D.G., Van Riemsdijk, W.H., and Koopal, L.K. (1995) Metal Ion Binding to Humic Substances: Application of the Non-Ideal Competitive Adsorption Model. Environmental Science & Technology 29: 446-457.

Benner, R. (2002) Chemical composition and reactivity. Biogeochemistry of Marine Dissolved Organic Matter 3: 56-63.

Berg, C., Vandieken, V., Thamdrup, B., and Jurgens, K. (2015) Significance of archaeal nitrification in hypoxic waters of the Baltic Sea. ISME J 9: 1319-1332.

Bernes, C. (2005) Change Beneath the Surface. An In-Depth Look at Sweden's Marine Environment. Fälth & Hässler, Värnamo, Sweden Monitor 19: 38-47.

Bernhard, A.E., and Kelly, J.J. (2016) Editorial: Linking Ecosystem Function to Microbial Diversity. Frontiers in Microbiology 7: 1041.

Bonaglia, S., Klawonn, I., De Brabandere, L., Deutsch, B., Thamdrup, B., and Brüchert, V. (2016) Denitrification and DNRA at the Baltic Sea oxic–anoxic interface: Substrate spectrum and kinetics. Limnology and Oceanography 61: 1900-1915.

Bonsdorff, E., Blomqvist, E.M., Mattila, J., and Norkko, A. (1997) Coastal eutrophication: Causes, consequences and perspectives in the Archipelago areas of the northern Baltic Sea. Estuarine, Coastal and Shelf Science 44: 63-72.

Boyd, E.S., and Peters, J.W. (2013) New insights into the evolutionary history of biological nitrogen fixation. Frontiers in Microbiology 4: 201.

Bradley, P.B., Sanderson, M.P., Frischer, M.E., Brofft, J., Booth, M.G., Kerkhof, L.J., and Bronk, D.A. (2010) Inorganic and organic nitrogen uptake by phytoplankton and heterotrophic bacteria in the stratified Mid-Atlantic Bight. Estuarine, Coastal and Shelf Science 88: 429-441.

Bristow, L.A., Sarode, N., Cartee, J., Caro-Quintero, A., Thamdrup, B., and Stewart, F.J. (2015) Biogeochemical and

31

metagenomic analysis of nitrite accumulation in the Gulf of Mexico hypoxic zone. Limnology and Oceanography 60: 1733-1750.

Bronk, D.A., See, J.H., Bradley, P., and Killberg, L. (2007) DON as a source of bioavailable nitrogen for phytoplankton. Biogeosciences 4: 283-296.

Burke, C., Steinberg, P., Rusch, D., Kjelleberg, S., and Thomas, T. (2011) Bacterial community assembly based on functional genes rather than species. Proceedings of the National Academy of Science of the United States of America 108: 14288-14293.

Codispoti, L.A., Brandes, J., P. Christensen, J., Devol, A., Naqvi, S.W.A., Paerl, H., and Yoshinari, T. (2000) The Oceanic Fixed Nitrogen and Nitrous Oxide Budgets: Moving Targets as We Enter the Anthropocene? Scientia Marina.

Crick, F.H. (1958) On protein synthesis. The Symposia of the Society for Experimental Biology 12: 138-163.

Daims, H., Lebedeva, E.V., Pjevac, P., Han, P., Herbold, C., Albertsen, M. et al. (2015) Complete nitrification by Nitrospira bacteria. Nature 528: 504-509.

Damashek, J., and Francis, C.A. (2018) Microbial Nitrogen Cycling in Estuaries: From Genes to Ecosystem Processes. Estuaries and Coasts 41: 626-660.

Dang, H., and Chen, C.-T.A. (2017) Ecological Energetic Perspectives on Responses of Nitrogen-Transforming Chemolithoautotrophic Microbiota to Changes in the Marine Environment. Frontiers in Microbiology 8: 1246.

Delmont, T.O., Quince, C., Shaiber, A., Esen, Ö.C., Lee, S.T.M., Rappé, M.S. et al. (2018) Nitrogen-fixing populations of Planctomycetes and Proteobacteria are abundant in surface ocean metagenomes. Nature Microbiology.

Deutsch, B., Alling, V., Humborg, C., Korth, F., and Mörth, C.M. (2012) Tracing inputs of terrestrial high molecular weight dissolved organic matter within the Baltic Sea ecosystem. Biogeosciences 9: 4465-4475.

Duce, R.A. (1986) The Impact of Atmospheric Nitrogen, Phosphorus, and Iron Species on Marine Biological Productivity. In The Role of Air-Sea Exchange in Geochemical Cycling. Buat-Ménard, P. (ed). Dordrecht: Springer Netherlands, pp. 497-529.

Erisman, J.W., Sutton, M.A., Galloway, J., Klimont, Z., and Winiwarter, W. (2008) How a century of ammonia synthesis changed the world. Nature Geoscience 1: 636.

Erwin, P.M., Pineda, M.C., Webster, N., Turon, X., and Lopez-Legentil, S. (2014) Down under the tunic: bacterial biodiversity hotspots and widespread ammonia-oxidizing archaea in coral reef ascidians. ISME J 8: 575-588.

Falkowski, P.G. (1997) Evolution of the nitrogen cycle and its influence on the biological sequestration of CO2 in the ocean. Nature 387: 272-275.

Falkowski, P.G., Fenchel, T., and Delong, E.F. (2008) The microbial engines that drive Earth's biogeochemical cycles. Science 320: 1034-1039.

Farnelid, H., Harder, J., Bentzon-Tilia, M., and Riemann, L. (2013a) Isolation of heterotrophic diazotrophic bacteria from estuarine surface waters. Environmental Microbiology 16: 3072-3082.

Farnelid, H., Bentzon-Tilia, M., Andersson, A.F., Bertilsson, S., Jost, G., Labrenz, M. et al. (2013b) Active nitrogen-fixing heterotrophic bacteria at and below the chemocline of the central Baltic Sea. ISME J 7: 1413-1423.

Farnelid, H., Andersson, A.F., Bertilsson, S., Al-Soud, W.A., Hansen, L.H., Sorensen, S. et al. (2011) Nitrogenase gene amplicons from global marine surface waters are dominated by genes of non-cyanobacteria. PLOS ONE 6: e19223.

Fenchel, T. (2008) The microbial loop – 25 years later. Journal of Experimental Marine Biology and Ecology 366: 99-103.

Finlay, B.J., Maberly, S.C., and Cooper, J.I. (1997) Microbial Diversity and Ecosystem Function. Oikos 80: 209-213.

Francis, C.A., Beman, J.M., and Kuypers, M.M. (2007) New processes and players in the nitrogen cycle: the microbial ecology of anaerobic and archaeal ammonia oxidation. ISME J 1: 19-27.

Francis, C.A., Roberts, K.J., Beman, J.M., Santoro, A.E., and Oakley, B.B. (2005) Ubiquity and diversity of ammonia-oxidizing archaea in water columns and sediments of the ocean. Proceedings of the National Academy of Sciences of the United States of America 102: 14683-14688.

32

Fuhrman, J.A. (2009) Microbial community structure and its functional implications. Nature 459: 193.

Galloway, J.N. (1998) The global nitrogen cycle: changes and consequences. Environmental Pollution 102: 15-24.

Galloway, J.N., and Cowling, E.B. (2002) Reactive nitrogen and the world: 200 years of change. Ambio 31: 64-71.

Galloway, J.N., Dentener, F.J., Capone, D.G., Boyer, E.W., Howarth, R.W., Seitzinger, S.P. et al. (2004) Nitrogen cycles: Past, present, and future. Biogeochemistry 70: 153-226.

Ganesh, S., Bristow, L.A., Larsen, M., Sarode, N., Thamdrup, B., and Stewart, F.J. (2015) Size-fraction partitioning of community gene transcription and nitrogen metabolism in a marine oxygen minimum zone. ISME J 9: 2682-2696.

Glibert, P.M., Harrison, J., Heil, C., and Seitzinger, S. (2006) Escalating Worldwide use of Urea – A Global Change Contributing to Coastal Eutrophication. Biogeochemistry 77: 441-463.

Glibert, P.M., Wilkerson, F.P., Dugdale, R.C., Raven, J.A., Dupont, C.L., Leavitt, P.R. et al. (2016) Pluses and minuses of ammonium and nitrate uptake and assimilation by phytoplankton and implications for productivity and community composition, with emphasis on nitrogen-enriched conditions. Limnology and Oceanography 61: 165-197.

Goldberg, I., Nadler, V., and Hochman, A. (1987) Mechanism of nitrogenase switch-off by oxygen. Journal of Bacteriology 169: 874-879.

Goldberg, S.J., Nelson, C.E., Viviani, D.A., Shulse, C.N., and Church, M.J. (2017) Cascading influence of inorganic nitrogen sources on DOM production, composition, lability and microbial community structure in the open ocean. Environmental Microbiology 19: 3450-3464.

Gran, V., and Pitkänen, H. (1999) Denitrification in estuarine sediments in the eastern Gulf of Finland, Baltic Sea. Hydrobiologia 393: 107-115.

Green, J.L., Bohannan, B.J.M., and Whitaker, R.J. (2008) Microbial Biogeography: From Taxonomy to Traits. Science 320: 1039-1043.

Gruber, N. (2008) Chapter 1 - The Marine Nitrogen Cycle: Overview and Challenges.

In Nitrogen in the Marine Environment (Second Edition). Capone, D.G., Bronk, D.A., Mulholland, M.R., and Carpenter, E.J. (eds). San Diego: Academic Press, pp. 1-50.

Gruber, N., and Galloway, J.N. (2008) An Earth-system perspective of the global nitrogen cycle. Nature 451: 293.

Haines, K.C., and Wheeler, P.A. (1978) Ammoniun and Nitrate Uptake by the Marine Macrophytes Hypnea Musvuformis (Rhodophyta) and Macrocystis Pyrifera (Phaeophyta) 2. Journal of Phycology 14: 319-324.

Haselkorn, R. (1978) Heterocysts. Annual Review of Plant Physiology 29: 319-344.

Hedges, J.I., Hansell, D.A., Carlson, C.A., Carlson, C.A., A Carlson, C., A Hansell, D. et al. (2002) Why Dissolved Organics Matter. Biogeochemistry of Marine Dissolved Organic Matter: 1-33.

HELCOM (2009) Eutrophication in the Baltic Sea – An integrated thematic assessment of the effects of nutrient enrichment and eutrophication in the Baltic Sea region. Baltic Sea Environment Proceedings 115B.

Hennecke, H., Kaluza, K., Thöny, B., Fuhrmann, M., Ludwig, W., and Stackebrandt, E. (1985) Concurrent evolution of nitrogenase genes and 16S rRNA in Rhizobium species and other nitrogen fixing bacteria. Archives of Microbiology 142: 342-348.

Herlemann, D.P., Labrenz, M., Jurgens, K., Bertilsson, S., Waniek, J.J., and Andersson, A.F. (2011) Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. ISME J 5: 1571-1579.

Herlemann, D.P.R., Lundin, D., Andersson, A.F., Labrenz, M., and Jürgens, K. (2016) Phylogenetic Signals of Salinity and Season in Bacterial Community Composition Across the Salinity Gradient of the Baltic Sea. Frontiers in Microbiology 7.

Heylen, K., Vanparys, B., Wittebolle, L., Verstraete, W., Boon, N., and De Vos, P. (2006) Cultivation of Denitrifying Bacteria: Optimization of Isolation Conditions and Diversity Study. Applied and Environmental Microbiology 72: 2637-2643.

Hietanen, S., and Lukkari, K. (2007) Effects of short-term anoxia on benthic denitrification, nutrient fluxes and phosphorus forms in coastal Baltic

33

sediment. Aquatic microbial ecology 49: 293-302.

Hietanen, S., Jäntti, H., Buizert, C., Jürgens, K., Labrenz, M., Voss, M., and Kuparinen, J. (2012) Hypoxia and nitrogen processing in the Baltic Sea water column. Limnology and Oceanography 57: 325-337.

Horz, H.P., Rotthauwe, J.H., Lukow, T., and Liesack, W. (2000) Identification of major subgroups of ammonia-oxidizing bacteria in environmental samples by T-RFLP analysis of amoA PCR products. Journal of Microbiological Methods 39: 197-204.

Hu, Y.O.O., Karlson, B., Charvet, S., and Andersson, A.F. (2016) Diversity of Pico- to Mesoplankton along the 2000 km Salinity Gradient of the Baltic Sea. Frontiers in Microbiology 7: 679.

Hugerth, L.W., Larsson, J., Alneberg, J., Lindh, M.V., Legrand, C., Pinhassi, J., and Andersson, A.F. (2015) Metagenome-assembled genomes uncover a global brackish microbiome. Genome Biology 16: 279.

Imhoff, J.F. (2016) New Dimensions in Microbial Ecology—Functional Genes in Studies to Unravel the Biodiversity and Role of Functional Microbial Groups in the Environment. Microorganisms 4: 19.

Jäntti, H., Ward, B.B., Dippner, J.W., and Hietanen, S. (2018) Nitrification and the ammonia-oxidizing communities in the central Baltic Sea water column. Estuarine, Coastal and Shelf Science 202: 280-289.

Jetten, M.S., Niftrik, L., Strous, M., Kartal, B., Keltjens, J.T., and Op den Camp, H.J. (2009) Biochemistry and molecular biology of anammox bacteria. Critical Reviews in Biochemistry and Molecular Biology 44: 65-84.

Jickells, T.D., Buitenhuis, E., Altieri, K., Baker, A.R., Capone, D., Duce, R.A. et al. (2017) A reevaluation of the magnitude and impacts of anthropogenic atmospheric nitrogen inputs on the ocean. Global Biogeochemical Cycles 31: 289-305.

Karl, D., Michaels, A., Bergman, B., Capone, D., Carpenter, E., Letelier, R. et al. (2002) Dinitrogen fixation in the world's oceans. Biogeochemistry 57: 47-98.

Karlson, K., Hulth, S., Ringdahl, K., and Rosenberg, R. (2005) Experimental recolonisation of Baltic Sea reduced sediments: survival of benthic macrofauna

and effects on nutrient cycling. Marine Ecology Progress Series 294: 35-49.

Kircher, M., and Kelso, J. (2010) High-throughput DNA sequencing--concepts and limitations. Bioessays 32: 524-536.

Kirshtein, J.D., Paerl, H.W., and Zehr, J.P. (1991) Amplification, Cloning, and Sequencing of a nifH Segment from Aquatic Microorganisms and Natural Communities. Applied and Environmental Microbiology 57: 2645-2650.

Kisand, V., and Wikner, J. (2003) Combining culture-dependent and -independent methodologies for estimation of richness of estuarine bacterioplankton consuming riverine dissolved organic matter. Applied and Environmental Microbiology 69: 3607-3616.

Kisand, V., Valente, A., Lahm, A., Tanet, G., and Lettieri, T. (2012) Phylogenetic and functional metagenomic profiling for assessing microbial biodiversity in environmental monitoring. PLOS ONE 7: e43630.

Kluyver, A.J., and Donker, H.J.L. (1926) Die Einheit der Biochemie. Chem Zelle Gewebe 13: 134-190.

Knudsen-Leerbeck, H., Mantikci, M., Bentzon-Tilia, M., Traving, S.J., Riemann, L., Hansen, J.L.S., and Markager, S. (2017) Seasonal dynamics and bioavailability of dissolved organic matter in two contrasting temperate estuaries. Biogeochemistry 134: 217-236.

Könneke, M., Bernhard, A.E., de la Torre, J.R., Walker, C.B., Waterbury, J.B., and Stahl, D.A. (2005) Isolation of an autotrophic ammonia-oxidizing marine archaeon. Nature 437: 543-546.

Krause, S., Le Roux, X., Niklaus, P.A., Van Bodegom, P.M., Lennon, J.T., Bertilsson, S. et al. (2014) Trait-based approaches for understanding microbial biodiversity and ecosystem functioning. Frontiers in Microbiology 5: 251.

Kujawinski, E.B., Longnecker, K., Barott, K.L., Weber, R.J.M., and Kido Soule, M.C. (2016) Microbial Community Structure Affects Marine Dissolved Organic Matter Composition. Frontiers in Marine Science 3: 45.

Kullenberg, G., and Jacobsen, T.S. (1981) The Baltic Sea: an outline of its physical

34

oceanography. Marine Pollution Bulletin 12: 183-186.

Kuypers, M.M.M., Marchant, H.K., and Kartal, B. (2018) The microbial nitrogen-cycling network. Nature Review Microbiology 16: 263-276.

Laas, P., Simm, J., Lips, I., Lips, U., Kisand, V., and Metsis, M. (2015) Redox-Specialized Bacterioplankton Metacommunity in a Temperate Estuary. PLOS ONE 10: e0122304.

Landa, M., Blain, S., Christaki, U., Monchy, S., and Obernosterer, I. (2015) Shifts in bacterial community composition associated with increased carbon cycling in a mosaic of phytoplankton blooms. ISME J 10: 39.

Larsson, U., Elmgren, R., and Wulff, F. (1985) Eutrophication and the Baltic Sea: Causes and Consequences.

Lawton, J.H., and Brown, V.K. (1994) Redundancy in Ecosystems. In Biodiversity and Ecosystem Function. Schulze, E.-D., and Mooney, H.A. (eds). Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 255-270.

Leininger, S., Urich, T., Schloter, M., Schwark, L., Qi, J., Nicol, G.W. et al. (2006) Archaea predominate among ammonia-oxidizing prokaryotes in soils. Nature 442: 806.

Lindh, M.V., Lefebure, R., Degerman, R., Lundin, D., Andersson, A., and Pinhassi, J. (2015) Consequences of increased terrestrial dissolved organic matter and temperature on bacterioplankton community composition during a Baltic Sea mesocosm experiment. Ambio 44 Suppl 3: 402-412.

Lindh, M.V., Sjöstedt, J., Casini, M., Andersson, A., Legrand, C., and Pinhassi, J. (2016) Local Environmental Conditions Shape Generalist But Not Specialist Components of Microbial Metacommunities in the Baltic Sea. Frontiers in Microbiology 7: 2078.

Louca, S., Parfrey, L.W., and Doebeli, M. (2016a) Decoupling function and taxonomy in the global ocean microbiome. Science 353: 1272-1277.

Louca, S., Jacques, S.M.S., Pires, A.P.F., Leal, J.S., Srivastava, D.S., Parfrey, L.W. et al. (2016b) High taxonomic variability despite stable functional structure across microbial communities. Nature Ecology & Evolution 1: 1-12.

Louca, S., Polz, M.F., Mazel, F., Albright, M.B.N., Huber, J.A., O’Connor, M.I. et al. (2018) Function and functional redundancy in microbial systems. Nature Ecology & Evolution 2: 936-943.

Lu, Z., Deng, Y., Van Nostrand, J.D., He, Z., Voordeckers, J., Zhou, A. et al. (2012) Microbial gene functions enriched in the Deepwater Horizon deep-sea oil plume. ISME J 6: 451-460.

McCarren, J., Becker, J.W., Repeta, D.J., Shi, Y., Young, C.R., Malmstrom, R.R. et al. (2010) Microbial community transcriptomes reveal microbes and metabolic pathways associated with dissolved organic matter turnover in the sea. Proceedings of the National Academy of Sciences of the United States of America 107: 16420-16427.

McElroy, M.B. (1983) Marine biological controls on atmospheric CO2 and climate. Nature 302: 328.

Moran, M.A., Satinsky, B., Gifford, S.M., Luo, H., Rivers, A., Chan, L.K. et al. (2013) Sizing up metatranscriptomics. ISME J 7: 237-243.

Moreno-Vivián, C., Cabello, P., Martínez-Luque, M., Blasco, R., and Castillo, F. (1999) Prokaryotic Nitrate Reduction: Molecular Properties and Functional Distinction among Bacterial Nitrate Reductases. Journal of Bacteriology 181: 6573-6584.

Mou, X., Sun, S., Edwards, R.A., Hodson, R.E., and Moran, M.A. (2008) Bacterial carbon processing by generalist species in the coastal ocean. Nature 451: 708-711.

Mulder, A., van de Graaf, A.A., Robertson, L.A., and Kuenen, J.G. (1995) Anaerobic ammonium oxidation discovered in a denitrifying fluidized bed reactor. FEMS Microbiology Ecology 16: 177-183.

Niemi, Å. (1979) Blue-green algal blooms and N:P ratios in the Baltic Sea. Acta botanica fennica 110: 57-61.

Ottesen, E.A., Young, C.R., Gifford, S.M., Eppley, J.M., Marin, R., Schuster, S.C. et al. (2014) Multispecies diel transcriptional oscillations in open ocean heterotrophic bacterial assemblages. Science 345: 207-212.

Pedros-Alio, C. (2006) Marine microbial diversity: can it be determined? Trends in Microbiology 14: 257-263.

Petersen, D.G., Blazewicz, S.J., Firestone, M., Herman, D.J., Turetsky, M., and Waldrop,

35

M. (2012) Abundance of microbial genes associated with nitrogen cycling as indices of biogeochemical process rates across a vegetation gradient in Alaska. Environmental Microbiology 14: 993-1008.

Pettersson, C., Allard, B., and Borén, H. (1997) River Discharge of Humic Substances and Humic-bound Metals to the Gulf of Bothnia. Estuarine, Coastal and Shelf Science 44: 533-541.

Peura, S., Sinclair, L., Bertilsson, S., and Eiler, A. (2015) Metagenomic insights into strategies of aerobic and anaerobic carbon and nitrogen transformation in boreal lakes. Scientific Reports 5: 12102.

Philippot, L. (2002) Denitrifying genes in bacterial and Archaeal genomes. Biochimica et Biophysica Acta (BBA) - Gene Structure and Expression 1577: 355-376.

Poretsky, R.S., Sun, S., Mou, X., and Moran, M.A. (2010) Transporter genes expressed by coastal bacterioplankton in response to dissolved organic carbon. Environmental Microbiology 12: 616-627.

Postgate, J.R. (1970) Biological Nitrogen Fixation. Nature 226: 25-27.

Prosser, J.I., and Nicol, G.W. (2008) Relative contributions of archaea and bacteria to aerobic ammonia oxidation in the environment. Environmental Microbiology 10: 2931-2941.

Rabalais, N.N. (2002) Nitrogen in aquatic ecosystems. Ambio 31: 102-112.

Reusch, T.B.H., Dierking, J., Andersson, H.C., Bonsdorff, E., Carstensen, J., Casini, M. et al. (2018) The Baltic Sea as a time machine for the future coastal ocean. Science Advances 4.

Richards, J., Sundermeier, T., Svetlanov, A., and Karzai, A.W. (2008) Quality control of bacterial mRNA decoding and decay. Biochimica et Biophysica Acta (BBA) - Gene Regulatory Mechanisms 1779: 574-582.

Richardson, D.J., and Watmough, N.J. (1999) Inorganic nitrogen metabolism in bacteria. Current Opinion in Chemical Biology 3: 207-219.

Rocca, J.D., Hall, E.K., Lennon, J.T., Evans, S.E., Waldrop, M.P., Cotner, J.B. et al. (2015) Relationships between protein-encoding gene abundance and corresponding process are commonly assumed yet rarely observed. ISME J 9: 1693-1699.

Rockström, J., Steffen, W., Noone, K., Persson, Å., Chapin Iii, F.S., Lambin, E.F. et al. (2009) A safe operating space for humanity. Nature 461: 472.

Rönnberg, C., and Bonsdorff, E. (2004) Baltic Sea eutrophication: area-specific ecological consequences. Hydrobiologia 514: 227-241.

Root, R.B. (1967) The Niche Exploitation Pattern of the Blue‐Gray Gnatcatcher. Ecological Monographs 37: 317-350.

Rotthauwe, J.H., Witzel, K.P., and Liesack, W. (1997) The ammonia monooxygenase structural gene amoA as a functional marker: Molecular fine-scale analysis of natural ammonia-oxidizing populations. Applied and Environmental Microbiology 63: 4704-4712.

Rowe, O.F., Dinasquet, J., Paczkowska, J., Figueroa, D., Riemann, L., and Andersson, A. (2018) Major differences in dissolved organic matter characteristics and bacterial processing over an extensive brackish water gradient, the Baltic Sea. Marine Chemistry.

Ryther, J.H., and Dunstan, W.M. (1971) Nitrogen, phosphorus, and eutrophication in the coastal marine environment. Science 171: 1008-1013.

Salles, J.F., Le Roux, X., and Poly, F. (2012) Relating Phylogenetic and Functional Diversity among Denitrifiers and Quantifying their Capacity to Predict Community Functioning. Frontiers in Microbiology 3: 209.

Schloesing, J.J., and Muntz, A. (1877) Sur la nitrification pas les ferments organisés. Comptes Rendus de l'Academie des Sciences 84: 301-303.

Seitzinger, S.P., and Sanders, R.W. (1997) Contribution of dissolved organic nitrogen from rivers to estuarine eutrophication. Marine Ecology Progress Series 159: 1-12.

Seitzinger, S.P., Harrison, J.A., Dumont, E., Beusen, A.H.W., and Bouwman, A.F. (2005) Sources and delivery of carbon, nitrogen, and phosphorus to the coastal zone: An overview of Global Nutrient Export from Watersheds (NEWS) models and their application. Global Biogeochemical Cycles 19.

Severin, I., Bentzon-Tilia, M., Moisander, P.H., and Riemann, L. (2015) Nitrogenase expression in estuarine bacterioplankton influenced by organic carbon and

36

availability of oxygen. FEMS Microbiology Letters.

Silvennoinen, H., Hietanen, S., Liikanen, A., Stange, C.F., Russow, R., Kuparinen, J., and Martikainen, P.J. (2007) Denitrification in the river estuaries of the northern Baltic Sea. Ambio 36: 134-140.

Singh, A., and Agrawal, M. (2008) Acid rain and its ecological consequences. Journal of Environmental Biology 29: 15-24.

Sipler, R.E., Kellogg, C.T.E., Connelly, T.L., Roberts, Q.N., Yager, P.L., and Bronk, D.A. (2017) Microbial Community Response to Terrestrially Derived Dissolved Organic Matter in the Coastal Arctic. Frontiers in Microbiology 8.

Sjöstedt, J., Pontarp, M., Tinta, T., Alfredsson, H., Turk, V., Lundberg, P. et al. (2013) Reduced diversity and changed bacterioplankton community composition do not affect utilization of dissolved organic matter in the Adriatic Sea. Aquatic Microbial Ecology 71: 15-24.

Smil, V. (2001) Enriching the earth: Fritz Haber, Carl Bosch, and the transformation of world food production: Cambridge: MIT Press.

Smith, J.M., Casciotti, K.L., Chavez, F.P., and Francis, C.A. (2014) Differential contributions of archaeal ammonia oxidizer ecotypes to nitrification in coastal surface waters. ISME J 8: 1704-1714.

Sogin, M.L., Morrison, H.G., Huber, J.A., Welch, D.M., Huse, S.M., Neal, P.R. et al. (2006) Microbial diversity in the deep sea and the underexplored “rare biosphere”. Proceedings of the National Academy of Sciences of the United States of America 103: 12115-12120.

Sriswasdi, S., Yang, C.-C., and Iwasaki, W. (2017) Generalist species drive microbial dispersion and evolution. Nature Communications 8: 1162.

Stal, L.J., Staal, M., and Villbrandt, M. (1999) Nutrient control of cyanobacterial blooms in the Baltic Sea. Aquatic Microbial Ecology 18: 165-173.

Stein, L.Y., and Klotz, M.G. (2016) The nitrogen cycle. Current Biology 26: R94-R98.

Stepanauskas, R., Jørgensen, N.O.G., Eigaard, O.R., Žvikas, A., Tranvik, L.J., and Leonardson, L. (2002) Summer inputs of riverine nutrients to the Baltic Sea: Bioavailability and eutrophication

relevance. Ecological Monographs 72: 579-597.

Sullivan, M.J., Gates, A.J., Appia-Ayme, C., Rowley, G., and Richardson, D.J. (2013) Copper control of bacterial nitrous oxide emission and its impact on vitamin B12-dependent metabolism. Proceedings of the National Academy of Science of the United States of America 110: 19926-19931.

Thureborn, P., Lundin, D., Plathan, J., Poole, A.M., Sjoberg, B.M., and Sjoling, S. (2013) A metagenomics transect into the deepest point of the Baltic Sea reveals clear stratification of microbial functional capacities. PLOS ONE 8: e74983.

Traving, S.J., Rowe, O., Jakobsen, N.M., Sorensen, H., Dinasquet, J., Stedmon, C.A. et al. (2017) The Effect of Increased Loads of Dissolved Organic Matter on Estuarine Microbial Community Composition and Function. Frontiers in Microbiology 8: 351.

Trenberth, K.E. (2011) Changes in precipitation with climate change. Climate Research 47: 123-138.

Treusch, A.H., Leininger, S., Kletzin, A., Schuster, S.C., Klenk, H.P., and Schleper, C. (2005) Novel genes for nitrite reductase and Amo-related proteins indicate a role of uncultivated mesophilic crenarchaeota in nitrogen cycling. Environmental Microbiology 7: 1985-1995.

Tyrrell, T. (1999) The relative influences of nitrogen and phosphorus on oceanic primary production. Nature 400: 525-531.

van Kessel, M.A., Speth, D.R., Albertsen, M., Nielsen, P.H., Op den Camp, H.J., Kartal, B. et al. (2015) Complete nitrification by a single microorganism. Nature 528: 555-559.

Venter, J.C., Remington, K., Heidelberg, J.F., Halpern, A.L., Rusch, D., Eisen, J.A. et al. (2004) Environmental Genome Shotgun Sequencing of the Sargasso Sea. Science 304: 66-74.

Vetterli, A., Hyytiäinen, K., Ahjos, M., Auvinen, P., Paulin, L., Hietanen, S., and Leskinen, E. (2015) Seasonal patterns of bacterial communities in the coastal brackish sediments of the Gulf of Finland, Baltic Sea. Estuarine, Coastal and Shelf Science 165: 86-96.

Veuger, B., Pitcher, A., Schouten, S., Sinninghe Damsté, J.S., and Middelburg, J.J. (2013) Nitrification and growth of autotrophic nitrifying bacteria and Thaumarchaeota in

37

the coastal North Sea. Biogeosciences 10: 1775-1785.

Vitousek, P.M., Hattenschwiler, S., Olander, L., and Allison, S. (2002) Nitrogen and nature. Ambio 31: 97-101.

Voss, M., Dippner, J.W., Humborg, C., Hürdler, J., Korth, F., Neumann, T. et al. (2011) History and scenarios of future development of Baltic Sea eutrophication. Estuarine, Coastal and Shelf Science 92: 307-322.

Walsh, E.A., Kirkpatrick, J.B., Rutherford, S.D., Smith, D.C., Sogin, M., and D'Hondt, S. (2015) Bacterial diversity and community composition from seasurface to subseafloor. ISME J 10: 979.

Wei, W., Isobe, K., Nishizawa, T., Zhu, L., Shiratori, Y., Ohte, N. et al. (2015) Higher diversity and abundance of denitrifying microorganisms in environments than considered previously. ISME J 9: 1954-1965.

Wheeler, P.A., and Kirchman, D.L. (1986) Utilization of inorganic and organic nitrogen by bacteria in marine systems1. Limnology and Oceanography 31: 998-1009.

Whitman, W.B., Coleman, D.C., and Wiebe, W.J. (1998) Prokaryotes: The unseen majority. Proceedings of the National Academy of Sciences of the United States of America 95: 6578-6583.

Wilson, P.W. (1957) On the sources of nitrogen of vegetation etc. Bacteriological Reviews 21: 215-226.

Wittebolle, L., Marzorati, M., Clement, L., Balloi, A., Daffonchio, D., Heylen, K. et al. (2009) Initial community evenness favours functionality under selective stress. Nature 458: 623.

Woese, C.R. (1987) Bacterial evolution. Microbiological Reviews 51: 221-271.

Woese, C.R., and Fox, G.E. (1977) Phylogenetic structure of the prokaryotic domain: The primary kingdoms. Proceedings of the National Academy of Sciences of the United States of America 74: 5088-5090.

Zehr, J.P., and McReynolds, L.A. (1989) Use of Degenerate Oligonucleotides for Amplification of the nifH Gene from the Marine Cyanobacterium Trichodesmium thiebautii. Applied and Environmental Microbiology 55: 2522-2526.

Zehr, J.P., Short, S.M., Jenkins, B.D., and Steward, G.F. (2003) Nitrogenase gene

diversity and microbial community structure: a cross-system comparison. Environmental Microbiology 5: 539-554

Zumft, W.G. (1997) Cell biology and molecular basis of denitrification. Microbiology and Molecular Biology Reviews 61: 533-616.

38

Paper I

39

40

Coupling biogeochemical process rates and metagenomic blueprints of coastal bacterial assemblages in the context of environmental change Trine Markussen1, Elisabeth M. Happel1, Jonna E. Teikari2, Vimala Huchaiah3, Johannes Alneberg4, Anders F. Andersson4, Kaarina Sivonen2, Lasse Riemann1, Mathias Middelboe1*, Veljo Kisand3* 1Marine Biological Section, Department of Biology, University of Copenhagen, Helsingør, Denmark

2Department of Microbiology, University of Helsinki, Helsinki, Finland

3Institute of Technology, University of Tartu, Tartu, Estonia

4KTH Royal Institute of Technology, Science for Life Laboratory, School of Biotechnology, Stockholm, Sweden *Correspondence: Mathias Middelboe: mmiddelboe@bio.ku.dk, Veljo Kisand: kisand@ut.ee Bacteria are major drivers of biogeochemical nutrient cycles and energy fluxes in marine environments, yet how bacterial communities respond to environmental change is not well known. Metagenomes allow examination of genetic responses of the entire microbial community to environmental change. However, it is challenging to link metagenomes directly to biogeochemical process rates. Here, we investigate metagenomic responses in natural bacterioplankton communities to simulated environmental stressors in the Baltic Sea, including increased river water input, increased nutrient concentration, and reduced oxygen level. This allowed us to identify informative prokaryotic gene markers, responding to environmental perturbation. Our results demonstrate that metagenomic and metabolic changes in bacterial communities in response to environmental stressors are influenced both by the initial community composition and by the biogeochemical factors shaping the functional response. Furthermore, the different sources of dissolved organic matter (DOM) had the largest impact on metagenomic blueprint. Most prominently, changes in DOM loads influenced specific transporter types reflecting the substrate availability and DOC assimilation and consumption pathways. The results provide new knowledge for developing models of ecosystem structure and biogeochemical cycling in future climate change scenarios and advance our exploration of the potential use of marine microorganisms as markers for environmental conditions.

Introduction

Prokaryotic microbial communities respond rapidly to environmental changes e.g. substrate composition, oxygen concentration, nutrient supply and salinity by changes in

their community composition and gene expression (Kirchman et al., 2004; Sjöstedt et al., 2012; Lindh et al., 2015). Consequently, the genetic composition of prokaryotic microbial communities is linked to concomitant environmental conditions, and changes in their combined gene-pool (genetic blueprint)

41

may potentially be used as an indicator of environmental changes. At the same time, prokaryotes are the main drivers of biogeochemical cycles, and changes in their abundance, activity and metabolism may therefore affect the prevalence of specific biogeochemical pathways (Fuhrman et al., 2015; Graham et al., 2016). These properties make prokaryotes an important, but so far largely overlooked group of organisms to include in environmental monitoring (Caruso et al., 2016). Thus, identification of informative prokaryotic gene markers changing in abundance in response to external forcing would be highly valuable in a monitoring context. Today, next generation sequencing techniques applied to environmental samples (e.g. metagenomics and metatranscriptomics) provide comprehensive and quantitative descriptions of the phylogenetic and functional diversity by using the total pool of nucleic acids (Gilbert and Dupont, 2011; Anantharaman et al., 2016). Collectively, metagenomes and metatranscriptomes can be used to identify couplings between environmental conditions and the microbial community composition and function. Likewise, recent studies used changes in metagenomes to assess the impact of anthropogenic stressors on microbial communities (Kisand et al., 2012; Won et al., 2017). Yet, linking such changes in the microbial blueprint with specific changes in biogeochemical processes still remains a major challenge.

Our understanding of microbial diversity in the Baltic Sea (BS) was recently expanded by

studies on bacterial biogeography (Herlemann et al., 2011) and seasonal community succession (Andersson et al., 2010; Herlemann et al., 2013). Further, the identification of key prokaryotic players and their functional traits in subsystems of the BS, e.g. surface waters and the oxic-anoxic interfaces of the deep basins (Labrenz et al., 2010; Grote et al., 2011; Farnelid et al., 2013), has made substantial progress in recent years. Recently, metagenomes from a time-series of sampling station in the central BS and cruises across the various basins were successfully assembled and used to establish a reference database (BARM – Baltic reference metagenome) (Alneberg et al., accepted), enabling analysis of any BS metagenome - or transcriptome sample by mapping the sequences onto the database and deducing the functional and/or taxonomic profile of the community. The advantage of specific reference database was demonstrated by resent study showing that microbial communities in the Baltic Sea, one of world’s largest brackish water bodies, represent clades not previously sequenced for other freshwater or marine environments (Hugherth et al 2015).

In this study, we explored metagenomic blueprints of natural microbial assemblages exposed to selected stressors related to major environmental concerns in the BS (e.g. reduced oxygen, cyanobacterial blooms, and increased input of riverine water). By performing experiments using seawater microcosms (10 L) from two sites in the BS representing different sub-basins, we sought

42

Table 1. In situ abundance and activity of bacterioplankton and viruses, concentrations of nutrients and environmental conditions. Mean values ± standard deviation for triplicate samples are shown. Nd, no data.

to link these blueprints to microbial abundance, productivity and activity measurements in order to quantify changes in the microbial metagenomes and activities in response to environmental perturbations. We linked effects of anthropogenic disturbances on bacterial communities and biogeochemical processes with the abundance of specific protein clusters and identified important DOC-transporting proteins in the metagenomic blueprint that were affected by specific stressors. Our analysis adds new insights to the complex influence of environmental changes on microbial community composition and function and emphasizes the challenge of identifying specific signature genes to be used as indicators of environmental status in marine monitoring.

Results

Characteristics of the investigated sites

The two investigated sites in the BS (Fig. S1) were characterized by the nutrient concentration, salinity and temperature (Table

1). The dissolved organic carbon (DOC) concentration was nearly 2 fold lower at Øresund site (Exp I) compared to Storfjärden site (Exp II), in accordance with the large input of humic DOC from terrestrial sources to the Gulf of Finland. As the dissolved organic nitrogen (DON) concentrations were similar at the two stations, the C:N ratios of the organic matter differed substantially between Øresund (C:N = 7.6) and Storfjärden (C:N = 13.3). Inorganic nutrient concentrations were generally low at both sites but with a higher inorganic N:P ratio at Øresund (N:P = 23.3) than at the Storfjärden site (N:P = 1.7). Bacterial abundance (BA) and production (BP) was 1.6-2 fold lower at the Øresund site in April compared to the Storfjärden site in July. The same trend was seen for viral abundances (VA) (Table 1, Table S1).

Abundance, production and activity of the bacterial community and viruses

Øresund (Exp I) Storfjärden (Exp II) Lapväärtti river Lielupe River

Sampling time April August April April

Bacterial abundance (× 106 cells mL-1) 1.8 ± 0.13 3.3 ± 0.26 nd nd

Bacterial production (µg C L-1 d-1) 1.95 ± 0.06 4.1 ± 0.3 nd nd

Viral abundance (× 106 viruses mL-1) 13.1 ± 1.0 15.5 ± 2.0 nd nd DOC (µM) 381 ± 7 667 ± 91 1561 ± 16 1337 ± 30 DON (µM) 50 ± 2 50 ± 2 78.0 ± 0.3 429 ± 16 NH4+ (µM) 0.24 ± 0.01 0.26 ± 0.01 1.82 ± 0.02 1.77 ± 0.06 NO3- (µM) 2.1 ± 0.2 0.44 ± 0.03 22.2 ± 0.5 548 ± 47 PO43- (µM) 0.09 ± 0.02 0.26 ± 0.03 0.37 ± 0.04 0.30 ± 0.07 Salinity 13.4 6.0 0 0 Temperature (°C) 7 14 nd nd

43

Microcosm experiments with filtered surface water (0.22 µm) were conducted at the two sites of the BS. Each microcosm received one of seven treatments (Fig. S2; DOMhum, addition of humic rich river water; DOMagri, addition of agricultural river water; DOMsea, addition of autochthonous plankton lysate; LTC, addition of Lysate of Toxic Cyanobacteria (Nodularia spumigena); NP, inorganic nutrient addition; LowO2, reduced O2 concentration; and control, no addition)). Microcosms were incubated at in situ temperature for three to four days.

In both experiments, an increase in cell abundance and production of the diluted natural bacterial communities was observed in all seven treatments (Fig. 1) with growth rates of 0.3 to 1.1 d-1, emphasizing the potential for shifts in the bacterial metagenomes in response to the treatments. DOMagri treatment caused elevated initial VA in Exp II, reflecting a high viral concentration in that river (Fig. 1). The VA remained relatively constant during the

Figure 1. Time course of bacterial abundance (BA, upper panel), viral particles abundance (VA, middle panel) and bacterial production (BP, bottom panel) in treatments for Exp I (left panel) and Exp II (right panel). DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous plankton lysate; LTC, lysate from toxic cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration; Control, controls. Significant difference between treatments and controls were tested using Dunnett’s test (Table S1).

44

incubation except for the LTC treatment in Exp I, where the increased BP supported a doubling of the VA during the stationary phase of the incubation.

Most pronounced bacterial growth was observed in the LTC treatment reflecting a large input of labile DOC in both experiments (Fig. 1 and S3). LTC treatments showed significant increase of protease (both

experiments), β-glucosidase (both experiments) and alkaline phosphatase activities (Exp I) compared to controls (Fig. 2, Dunnett’s test; Table S2). DOMagri treatments significantly increased BA (both experiments) and BP (Exp II), VA (both experiments), protease (both experiments), β-glucosidase (Exp II) and alkaline phosphatase activities (both experiments, Fig. 1 and 2) compared with the controls (Table S2).

Figure 2. Extracellular enzymatic activities in microcosms from Exp I (left panel) and Exp II (right panel). Protease activities are shown in upper panel, β-glucosidase activity in middle panel and alkaline phosphatase activity in the bottom panel. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous plankton lysate; LTC, lysate from toxic cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration; Control, controls. Significant difference between treatments and controls were tested using Dunnett’s test (Table S1).

45

Treatments Number of

significant COGsa (% of all)

Total counts of significant COGs

(cpm)d

Number of significant COGs used in downstream analysis

Up- represented COGs

Down- represented COGs

%DOC-related COGs of up- represented

Exp I DOMhum 207 (10%) 120226 207a 132 75 36%

DOMagri 16 (1%) 6380 16a 14 2 21%

DOMsea 23 (1%) 18382 23a 20 3 65%

LTC 1480 (72%) 762058 167 (8%)b 57 108 37%

NP 15 (1%) 6919 15a 14 1 21%

LowO2 40 (2%) 7912 40a 22 18 18%

Exp II DOMhum 361 (18%) 188767 83 (4%)c 56 27 46%

DOMagri 368 (18%) 151391 166 (8%)c 43 123 40%

DOMsea 268 (13%) 132625 90 (4%)c 29 61 52%

LTC 1525 (76%) 795492 145 (7%)b 47 98 40%

NP 390 (19%) 173299 95 (4%)c 21 74 33%

LowO2 13 (1%) 6843 13a 10 3 30%

Table 2. Identification of COGs up- or down-represented in response to treatment. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous plankton lysate; LTC, lysate from toxic cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration. Extraction of significant COGs was performed with different filtering criteria; a Criteria for extracted significant COGs were p < 0.01, FDR < 0.05. b Criteria for extracted strong and significant COGs were p < 0.01, FDR < 0.05, logFC > 1.5. c Criteria for extracted strong and significant COGs were p < 0.01, FDR < 0.05, logFC > 0.4. d Average of normalized counts (count-per-million, cpm) of triplicates samples for all significant COGs (p < 0.01, FDR < 0.05). NF indicates that no additional filter was applied on treatment.

DOMhum treatment had no effect on bacterial growth in Exp I, but significantly stimulated BA and production in Exp II (Table S2). DOMsea treatment significantly increased BA (Exp I), BP (Exp II), protease activity (Exp I) and alkaline phosphatase activity (both experiments) compared to the controls (Fig. 1 and 2, Table S2). Treatments with NP and LowO2 did not affect BA or BP relative to the controls (both experiments, Table S2). The dynamics of inorganic nutrient concentrations differed between the two experiments, but

also between treatments (Fig. S4). DOMagri treatment constituted a significant input of NO3 in Exp I but not in Exp II (Table S2). The high DOC:DON and low inorganic N:P ratios at the Exp II site was associated with a significant decrease in ammonium concentration in the treatments amended with inorganic N (DOMagri, DOMhum and NP). This was not the case at the Exp I site (high N:P ratio), where the phosphate concentration was reduced to below detection limit within 40 h.

46

Community composition at the end of experiments

At the end of experiments (Tend) microbial community composition differed between Exp I and Exp II and higher phylogenetic variability was observed in Exp II compared to Exp I (Fig. 3A and B). Actinobacteria were more abundant in Exp II compared to Exp I (median Exp I: 0.7%; median Exp II: 3.3%) whereas Gammaproteobacteria were both the most dominant sub-phylum in Exp I and more abundant compared to Exp II (median Exp I: 43.6%; median Exp II 9.4%). In both experiments, the abundance of Betaproteobacteria was stimulated in the DOMhum and DOMsea treatments relative to the controls. Addition of LTC changed the abundance of several sub-phyla compared with controls including a large increase in Gammaproteobacteria and a decrease in Actinobacteria and Betaproteobacteria. The taxonomic richness (calculated as abundance-based coverage estimator (ACE)) was significantly higher in Exp II compared to Exp I (R2 = 0.24, p < 0.001), however, no significant differences in richness were observed between treatments from same experiment (Fig. S5A). Furthermore, Shannon diversity was significantly higher in bacterial communities for Exp II than Exp I (R2 = 0.67, p < 0.001, Fig. S5B) and only LTC bacterial communities (Exp I and II) showed significantly lower diversity compared to controls (p < 0.05).

Functional metagenomic profile at the end of experiments

Metagenomic sequencing of the total community DNA from Tend resulted in 2.6 to 19.1 × 106 raw reads in Exp I and 5.0 to 12.4 × 106 raw reads in Exp II per sample. The number of reads that mapped to the BARM database varied between 1.1 and 3.9 × 106 for Exp I samples, and between 1.2 and 3.5 × 106 for Exp II samples, corresponding to 30-70% of the total number reads. Mapped reads were annotated using Cluster of Orthologous Groups (COGs). Multidimensional scaling (MDS) plot of functional community profiles based on distances corresponding to log2-fold-changes (logFC) between each sample showed that the treatments clustered according to experimental site (Fig. S6). However, COGs associated with distinct metabolic pathways varied between locations. For example, we observed that Exp I had higher abundance of the key enzyme 2-keto-3-deoxy-6-phosphogluconate aldolase genes (COG0800) in the Entner–Doudoroff (ED) pathway than Exp II (p < 0.01, logFC = 1.29), while Exp II treatments had higher abundance of the key enzyme 6-phosphofructokinase genes (COG0205, p < 0.01, logFC = 0.69) involved in the common glycolysis (Embden–Meyerhof–Parnas - EMP) pathway compared to Exp I (data not shown).

Effects of treatment on metagenomic blueprints

In order to identify COGs which were representative for specific treatments, we compared treatments with controls from the given site. The number of significantly

47

Figure 3. Microbial community composition assessed by 16S rRNA gene sequence analysis of treatments from Exp I and treatments from Exp II at time point Tend (T78 for Exp I and T96 for Exp II). Major taxonomic groups at Pylum/Sub-Phylum level (A) and Order level (B).Clustering of COGs significantly up-represented in one or more treatments compared to controls for (C) Exp I (n = 221) and (D) Exp II (n = 144). Heatmap coloring reflects the Z score of the normalized abundances (cpm) of each COG across all clustered data. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous plankton lysate; LTC, lysate from toxic cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration; Control, controls.

48

responding genes was higher for treatments in Exp II than in Exp I (p < 0.01, FDR > 0.05, Table 2). Especially, the LTC treatments affected a high number of COGs (72% of all annotated COGs). In order to shortlist the number of COGs for more detailed analysis of metabolic pathways, an additional logFC filter was therefore applied for the statistical analysis of selected treatments (see Experimental procedures, Table 2) resulting in a total of 468 and 592 responding COGs selected for further analysis for Exp I and Exp II, respectively. The up-represented COGs accounted for 55% and 35% of total responding COGs for Exp I and Exp II, respectively (Table S3).

In both experiments we observed an overall dominance of responding COGs related to metabolism and transport (Fig. S7) with DOC-related COGs accounting for 18-65% (Exp I) and 30-52% (Exp II) of all up-represented COGs in the different treatments (Table 2, Table S4). In addition, some of the most up-represented COGs relative to the control treatments were from the categories 'Transcription' and 'Signal transduction mechanisms' with functions related to iron metabolism (COG1321), phage transcription (COG3311), sporulation (COG1774, COG3854), denitrification or amino acid/polyamine metabolism and transport (COG1348, COG4566, COG3284, COG2909) (Table S2). Further, in several treatments (DOMhum, DOMagri and NP in Exp I, and DOMhum and LowO2 in Exp II), a relatively large fraction of the up-represented COGs belonged to the 'Mobilome: prophages, transposons' category (Fig. S7).

Cluster analysis of up-represented COGs unique to or shared between treatments demonstrated treatment-related division of functional properties (Fig. 3C and D). A k-medoids clustering analysis to partition the COGs into five clusters based on similarities in abundance profiles, identified distinct clusters of COGs associated with specific environmental stressors (Fig. S8, Table S3). For example, cluster 4 were dominated by COGs from the DOMhum treatment (Exp I), and COGs from cluster 5 (Exp I) were only up-represented in the LowO2 treatment (Fig. S8A, Table S3). The clustering further showed functional similarities between treatments, such as COGs in Cluster 3 (Exp II) which were up-represented in DOMhum, DOMagri and NP treatments (Fig. S8B, Table S3).

Distinct categories of DOC-related COGs were affected in the different treatments (Fig. 4, Table S4). In general, the most up-represented category in Exp I was ‘Carbohydrate transport and metabolism’ (19% of all up-represented DOC-related COGs). Especially in the DOMhum, DOMsea and LTC treatments, the COG category 'Carbohydrate transport and metabolism' was highly up-represented relative to the control. In DOMhum, this was combined with COGs involved in phosphate transport, whereas in DOMsea, also COGs assigned to 'Inorganic ion and transport metabolism' (23%) and proteases responded to the treatment (Table S4). A large variety of COGs involved in carbohydrate metabolism was up-represented in the LTC treatment, including COGs associated with the metabolism of simple sugars (e.g. fructose, sucrose, mannose and starch). The

49

Figu

re 4

. In

crea

se (

as l

ogFC

) in

abu

ndan

ce o

f si

gnifi

cant

ly c

hang

ing

tran

spor

ter

CO

Gs

com

pare

d to

con

trol

s cl

assi

fied

into

tr

ansp

orte

r typ

e (le

ft pa

nel),

com

poun

ds in

volv

ed in

act

ivity

of t

hese

tran

spor

ters

(mid

dle

pane

l), a

nd m

ajor

d C

OF

cate

gori

es (r

ight

pa

nel).

DO

Mhu

m, f

resh

wat

er fr

om h

umic

rich

rive

r; D

OM

agri

, fre

sh w

ater

from

pol

lute

d ri

ver i

nflu

ence

d by

agr

icul

ture

; DO

Mse

a,

auto

chth

onou

s pl

ankt

on ly

sate

; LTC

, lys

ate

from

toxi

c cy

anob

acte

rial

cul

ture

; NP,

inor

gani

c nu

trie

nt a

dditi

on; L

owO

2, re

duce

d O

2 co

ncen

trat

ion.

50

metagenomic response for NP and DOMagri treatments was highly similar, sharing 71% of all up-represented COGs and with a limited number of up-represented DOC-related COGs (n = 4 for DOMagri and n = 3 for NP). Likewise, no dominant COG category was affected by the LowO2 treatment and the affected DOC-related COGs (n = 4) showed less than 0.7 logFC difference in abundance compared to controls (Table S4).

In contrast to Exp I, the most affected category in Exp II was 'Inorganic ion and transport metabolism' (17% of all up-represented DOC-related COGs). The DOMhum treatment in Exp II affected COGs involved in nucleotide and amino acid metabolism pathways (23%) rather than carbohydrate metabolism (12%). The response to DOMagri treatment in Exp II was also different from Exp I, showing high abundance of DOC-related COGs in the category 'Nucleotide metabolism and transport' (24%) and with 60% overlap with DOMhum in responding COGs (Fig. 3D). Both treatments showed up-representation of enzymes associated with nitrogen metabolism (Table S3). As in Exp I, COGs associated with carbohydrate metabolism was highly up-represented in the DOMsea and LTC treatment (Table S3). In treatments amended with various levels of inorganic N (DOMagri, DOMhum and NP) up-representation of COGs related to purine degradation was observed. A limited number of affected DOC-related COGs changed in abundance in LowO2 (n = 3).

Comparison of the treatment-specific metagenomes between the two experiments identified a common set of COGs that responded similarly to the imposed

treatments, thus representing potential indicator COGs for the specific stressors (Fig. S9). Such common signature COGs with higher relative abundance compared to controls were particular prevalent in DOMhum and LTC treatments. Shared COGs between DOMhum treatments involved COGs belonging to the 'Mobilome: prophages, transposons' category and five COGs involved in carbohydrate metabolism including three tripartite ATP-independent periplasmic (TRAP) transporters. Shared COGs between LTC treatments (logFC > 1.5) included two COGs assigned as ABC transporters, and COGs involved in carbohydrate metabolism and amino acid metabolism. The shared COGs in the DOMsea treatments also included DOC-related COGs (Fig. S9). No COGs were shared between LowO2 treatments.

Changes in abundance of transporters

Changes in abundance of specific transporter genes were analyzed in the subset of significantly up-represented COGs (logFC > 0, FDR < 0.05, p-adjusted < 0.001, see Experimental Procedures). In Exp I a number of important transporters responded to different treatments (Fig. 4). Both in the DOMhum treatment (ABC type and three TRAP type transporters) and in the DOMsea treatment (tonB-dependent transporters) specific transporters increased significantly in abundance compared with the controls. No transporters increased in abundance in DOMagri, NP and LowO2 treatments compared to control, whereas many transporters were identified in the LTC treatment, including one ABC-type transporter, seven permeases, one antiporter, two tonB-dependent transporters and three

51

symporters. In total 23 transporter genes increased in abundance in response to the LTC treatment. The diversity of transporters was higher in Exp II (Fig. 4) compared to Exp I, with the most pronounced response in treatments receiving river water: 33 transporters responded with increasing counts in DOMhum treatment, including 19 ABC-transporters and five TRAP-transporters, and 13 transporters in DOMagri, including three ABC-transporters. LTC treatment was not dominated by any specific type of transporters (n = 4).

Linking activity and function: relationship between COG abundance and microbial activity

Detrended correspondence analysis (DCA) of COG abundances responding significantly to the treatments and linear fitting of response variables identified significant correlations (p < 0.05) between cluster of COGs and activity/abundance/concentration variables. The linear fitting indicated that elevated BP, extracellular enzymatic activities and amino acids uptake were strongly associated to LTC treatment (data not shown) suggesting that the strong response from adding high loads of DOC from cyanobacterial lysate shaded the response from the remaining treatments. In order to analyze the relationships in the rest of the treatments, the subset of COG and activity-abundance-nutrient variables from control, DOMhum, DOMagri, NP and LowO2 were analyzed independently (Fig. 5, Table S5). In Exp I three independent clusters were observed (Fig. 5A). The first cluster of COGs was associated with protease activity (co-

related with alkaline phosphatase activity and PO4 concentration) and characterized by COGs belonging to 'Carbohydrate transport and metabolism' and 'General function' categories. The second cluster of COGs related to amino acid uptake (co-related to DOC and ammonium concentration) and was enriched with 'Mobilome: prophages, transposons', primarily phage related proteins (Table S4). Other strongly associated COGs were involved in pyrimidine metabolism and amino acid transport. The third cluster of COGs was strongly associated to BP, containing mostly COG associated with unknown or general functions, except included three ABC transporters and COGs related to carbohydrate metabolism pathways (Table S5).

In Exp II, COG clusters and associated variables were distinct from the association analysis of Exp I (Fig. 5B, Table S4). One cluster of COGs associated with alkaline phosphatase activity and was enriched with COGs belonging to the 'General function' category including two ABC-type transporters. The COGs associated with protease activity belonged to 'Carbohydrate transport and metabolism' and 'Function unknown', and included two proteases. The third cluster of COGs which was strongly associated with amino acid uptake, BA and BP contained COGs belonging to the category 'Inorganic ion transport and metabolism'. Other strongly related COGs in this cluster were enzymes involved in purine degradation and in nitrate assimilation. β-glucosidase activity was significantly related to DCA

52

Figure 5. Association based on detrended correspondence analysis between relative abundance of reads in COGs and measured activities, rates and concentrations of nutrients in: (A) Exp I: BP (r2 = 0.2518, p = 0.088), DOC (r2 = 0.42, p = 0.007), Ammonium (r2 = 0.38, p = 0.019), PO4 (r2 = 0.37, p = 0.095), protease activity (r2 = 0.50, p = 0.004), alkaline phosphatase (r2 = 0.34, p = 0.026). BA, VA, DON, NO3 and β-glucosidase were not significant. (B) Exp II: BP (r2 = 0.85, p = 0.001), BA (r2 = 0.62, p = 0.002), VA (r2= 0.33, p = 0.027), DOC (r2 = 0.26, p = 0.086), Protease activity (r2 = 0.79, p = 0.001), β-glucosidase (r2 = 0.69, p = 0.001), alkaline phosphatase (r2 = 0.42, p = 0.010). DON, Ammonium, PO4 and NO3 were not significant. Pie charts indicate proportions of COG clustered into COG categories which are related with measured activities, rates and concentrations of nutrients.

gradients (r2 = 0.69, p = 0.001) but was not associated with any COG clusters.

Discussion

Prokaryotic communities are capable of adjusting to the physical and chemical factors defining a specific environment. Consequently, functional differences in microbial communities can be used to discriminate qualitatively dissimilar environments (Tringe et al., 2005; Dinsdale et al., 2008; Gianoulis et al., 2009). Indeed, our results document selection for unique functional metagenomic patterns in marine microbial communities in response to selected environmental stressors at two geographic sites. We were further able to link the changes in metagenomic profiles to dynamics in microbial activities caused by environmental perturbations. This shows how adaptations to environmental changes can shape both microbial community structure and functional metagenomic profiles over a short time scale.

Site specific blueprint

The major differences in key metabolic pathways at the two sampling sites was in agreement with previous observations of elevated abundance of Entner–Doudoroff (ED) and Embden–Meyerhof–Parnas (EMP) pathway genes in higher and lower salinity environments, respectively (Dupont et al.,

53

2014; Fortunato and Crump, 2015). This was further supported by the high abundances of Gammaproteobacteria in Exp I (high salinity) and high abundances of Actinobacteria in Exp II (low salinity), as these bacterial groups previously have been found to be associated with the ED and EMP pathway, respectively (Flamholz et al., 2013; Fortunato and Crump, 2015). The observed relationship between salinity, microbial community composition and metabolic function suggested that salinity was an important environmental factor shaping the metagenomic blueprint at the two sites, as also previously shown for microbial communities in the BS (Herlemann et al., 2011; Dupont et al., 2014).

In general, the number and composition of COGs that changed due Embden–Meyerhof–Parnas to manipulations differed between the two sites suggesting that the distinct communities showed different adaptations to the same environmental stressors. However, the distinct responses may also reflect differences in other environmental parameters such as temperature and seasonal variations in the composition of autochthonous and allochthonous DOM added to the experiments.

Changes in metagenomes in response to treatments

The significant changes in the metagenomes associated with the applied environmental stressors support previous suggestions that different sources of bioavailable DOC induce different genomic and functional responses in a microbial community (McCarren et al., 2010; Poretsky et al., 2010). Our study indicates several links between changes in abundance of specific groups of functional genes and differences in community activity and

environmental parameters. For example, the elevated C:N ratio resulting from DOM input from the humic river in Exp I was associated with an increase in COGs related to carbohydrate metabolism, indicating the processing of new sources of bioavailable DOC. Furthermore, increased abundance of COGs involved in phosphate transport suggested that the input of humic matter stimulated phosphate uptake as previously suggested (Carlsson et al., 1993). The shared COGs in the DOMhum treatments belonged mainly to TRAP transporters and carbohydrate metabolism pathways, demonstrating that the added humic DOM increased the abundance of COGs involved in specific DOC assimilation and consumption pathways in both experiments. TRAP-type transporters transfer low molecular weight extracellular solutes (Forward et al., 1997), especially organic acids (Mulligan et al., 2011) into the cell. These compounds are likely products of extracellular hydrolysis of high molecular weight humic material, and may therefore be indicators of the input of humic DOM with river water.

Interestingly, we observed that DOMagri and NP treatments both resulted in up-representation of COGs that correlated with amino acid uptake, NH4 and DOC concentration (Table S3 and S5), suggesting that the increased nitrate and ammonium loads shaped similar genetic responses related to N-metabolism in these treatments. In Exp II, on the other hand, similar change in metagenome was observed in DOMagri and DOMhum, characterized by a stimulation of DOC transporters (e.g. particularly TRAP and ABC transporters). This suggested that the bacterial community mainly responded to the

54

DOC input from the river water. The simultaneous net uptake of inorganic N in these treatments corresponded with high C:N ratio and N limitation and was reflected in increased abundance of COGs involved in nitrogen fixation. Increased prevalence of COGs related to proteases in the DOMsea treatments (i.e. the top three most up-represented COGs) was in agreement with the increased protease activity, suggesting that the DOM from the natural plankton lysate stimulated abundance of organims with active protein metabolism. Natural plankton lysate represents a variety of organic compounds that are readily recycled by heterotrophs (Amon and Benner, 1996) as indicated by the elevated protease activity. Likewise, LTC treatments showed the largest effect on bacterial metagenomic blueprints in both experiments, with dominance of COGs associated with general metabolic functions. This was also reflected in the highly elevated microbial activity and abundance associated with the significant increase in bioavailable DOM. Cell lysates are rich in proteins and DNA and can constitute a high quality substrate source for heterotrophic bacteria (Middelboe and Jørgensen, 2006; Haaber and Middelboe, 2009), as also indicated by the elevated protease, glucosidase and phosphatase activity in these treatments. This change of metabolic properties was also associated with shifts in microbial community composition in both LTC experiments towards dominance of fast-growing Gammaproteobacteria (Eilers et al., 2000; Pinhassi and Berman, 2003) known for their ability to exploit DOM when available (Fuchs et al., 2000; Herlemann et al., 2014).

The LowO2 treatment showed generally few metagenomic changes relative to the control in both experiments, in agreement with the similar development in bacterial abundance and activities (e.g. BA, BP, and enzymatic activities) and low or unchanged DOC consumption measured for this treatment compared to control. Consequently, within the experimental time frame, the substrate lability and composition (e.g. C:N:P ratio, humic components and protein content) had a much more pronounced effect on the bacterial community than reducing oxygen concentration to 15% saturation.

Some of the most up-represented COGs in the treatments compared to controls involved regulatory mechanisms such as transcriptional regulators. These were especially observed in the DOM treatments, e.g. DOMhum, DOMagri and LTC, indicating an important role of these COGs for rapid adaptations to changing substrate conditions (Cases et al., 2003). Likewise, several phage related proteins and transposases were up-represented in these treatments compared to controls. The function of these COGs may be associated with horizontal gene transfer, a common source of microbial diversity and genomic plasticity, which is often enriched in highly dynamic environments (Angly et al., 2009). Alternatively, the up-representation may reflect increased induction of prophages following exposure to changing environmental conditions or by stimulation of bacterial metabolic activity.

Effects of environmental changes on bacterial transporters

In order to successfully attain essential nutrients and DOM from marine environment

55

with limited resources, bacteria need efficient transport systems. Bacterial transporters are frequently observed in microbial metatranscriptome and metaproteome data and can be important indicators of the uptake of organic and inorganic compounds used by the bacteria (McCarren et al., 2010; Morris et al., 2010; Poretsky et al., 2010; Bergauer et al., 2018). Therefore, we analyzed abundance change of various transporter types in the metagenomes, grouping the most abundant transporters using COGs annotation. The large increase of transporter abundance associated with LTC, DOMhum and DOMsea treatments in Exp I indicated that transporters were particularly responsive to input of organic matter rather than inorganic nutrients. In Exp II on the other hand, DOM additions stimulated a variety of different transporters that in some cases were specific for the specific DOM sources, e.g. an increased abundance of ABC-transporters associated with uptake of complex carbohydrates and N-compounds in the DOMhum treatment (Fig. 4). This corresponds well with previous indications that certain ABC-type transporters are responsible for uptake of possibly refractory DOM in the BS (Jiao and Zheng, 2011) and such transporters may therefore be useful as indicators of increased load of refractory material. DOMsea treatments from both Exp I and II showed up-represented TonB-dependent transporters, which are involved in bacterial uptake of scarce resources such as iron-complexes (e.g. siderophores, heme, etc.), vitamin B12, and some carbohydrates from nutrient-limiting environments (Tang et al., 2012). These transporters are often associated with Flavobacteria (Teeling, 2012), which were also dominant in our experiments (data not shown).

Linking activity with genetic blueprints

We found significant correlations between up-represented COGs and the microbial and chemical parameters in the DOMhum, DOMagri, DOMsea, LowO2 and NP treatments (Fig. 5, Table S5) allowing us to associate the change in metagenomes to biochemical processes at various environmental conditions. Most prominently alkaline phosphatase and protease activity correlated with several COGs involved in metabolic pathways responsible for transport and metabolism of inorganic compounds, carbohydrates, and organic N-compounds (i.e. amino acids). Amino acid uptake correlated with several COGs involved in nitrogen metabolism pathways, in particular purine degradation and pyrimidine biosynthesis. These COGs were found mainly in DOMhum and DOMagri treatments suggesting higher availability of purines and pyrimidines in the river water.

In conclusion, we demonstrated that bacterioplankton communities responded significantly to the environmental disturbances in terms of activity and gene abundance. Notably, the metagenomic and metabolic changes in bacterial communities could be ranked in response to various stressors. The most prominent response occurred to changes in DOM load from river water, and depended on the catchment area, i.e. humic and agricultural influenced catchments. Cyanobacterial lysates affected the microbial activity and metagenome mainly via an increased load of labile organic compounds leading to increased turnover of organic compounds and remineralization. Moderate decrease in oxygen concentration

56

did not affect the metabolism or genetic composition of bacterial communities within the relatively short time frame of the experiments (3-4 days). The complex and diverse effects of the experimental manipulations on the bacterial metagenome and metabolic activity highlight the challenge of linking specific metagenomic changes to specific environmental conditions in the BS. Nonetheless, our experiments allowed us to identify key characteristics of metagenomic profiles in response to environmental changes and link these to process rates. This provides a first step in exploring the use of bacterial gene abundance as indicators of biogeochemical processes and environmental conditions.

Experimental procedures

Study sites

Seawater (300 L) was collected from 5 m deph in Øresund (56°3'26" N 12°38'45" E; Exp I) on 20th of April 2015 and Storfjärden (Gulf of Finland; 59°51'12" N 23°16'19" E; Exp II) 27th of July 2015. Both sites are characterized by moderate to low salinity (13.4 and 6.0 for Øresund and Storfjärden, respectively) and in situ temperature was 7°C in April and 14°C in July (Table 1). River water was collected 2 days before the experiments from the humic influenced Lapväärtti river (62°14'21" N 21°34'38" E, Finland) and the Lielupe river (56°48'42" N 23°35'5" E, Latvia) which is mainly influenced by agricultural runoff (Fig. S1). River water was filtered through 0.22 µm capsule filters (Optical XL filter, Millipore) and kept in the dark at 4°C until use. The salinity of the river water was adjusted to Øresund or Storfjärden in situ concentrations with muffled NaCl.

Experimental design

Seawater was transported in acid-washed Milli-Q rinsed polycarbonate bottles to a climate room with in situ temperature within 1 h. The seawater was filtered in two ways (Fig. S2): (i) to prepare inoculum for the batch incubations, water was sieved through a 10 µm plankton net; (ii) for generating bacteria free water, which was obtained by additionally filtering water through 0.22 µm capsule filters (Optical XL filter, Millipore), and subsequently served as medium for the treatments. Recovered material collected on the 10 µm plankton net during filtration step (i) (from 60 L) was collected and resuspended in 50 ml 0.22 µm filtered seawater, sonicated and sterile filtered with a 0.22 µm Sterivex capsule filter (Millipore) and used as autochthonous DOM source for DOMsea treatments. Cyanobacterial lysate was prepared by freezing and thawing cultivated Nodularia spumigena UHCC 0039 cells (University of Helsinki Culture Collection, HAMBI). Lysate was 0.22 µm filtered prior to use.

The seven treatments were prepared and inoculated as followed (Fig. S2): 1) DOMhum; addition of 20% 0.22 µm filtered humic rich river water from Lapväärtti, 2) DOMagri; addition of 20% 0.22 µm filtered agricultural river water from Lielupe, 3) DOMsea; addition of autochthonous DOM source, 4) LTC; addition of lysate from toxic cyanobacteria, 5) NP; addition of inorganic nutrients to final concentration of 1 µM phosphate, 1 µM ammonia and 9 µM nitrate, 6) LowO2; oxygen concentration reduced to ~15% saturation by bubbling with N2-gas, and 7) control; no manipulation. The 10 L microcosms (six

57

treatments in triplicates and controls in six replicates) containing bacteria-free water were inoculated with bacteria at a ratio of 1:5 and incubated in the dark in a climate room at in situ temperature for 3-4 days. Mixing was provided by continuous bubbling with atmospheric air (with the exception of LowO2 treatments). In situ collected samples and the manipulated batch treatments were initially sub-sampled for measurements of bacterial abundance (BA) and production (BP), O2 consumption, inorganic nutrient and organic matter concentrations (DOC/DON), amino acid uptake and activity of bacterial extracellular enzymes at time zero, and subsequently with 6 to 12 h intervals in the batch experiments. For analysis of bacterial community composition, samples were collected from the last time point (Tend, 78 h for Exp I and 96 h for Exp II) on 0.22 µm filters for 16S rRNA gene fragment sequencing and for total community DNA based metagenomics. Full details on bacterial and viral abundance (VA), bacterial production (BP), amino acid uptake, DOC/DON concentrations, oxygen saturation, extracellular enzymes activity and chemical analyses are provided in Experimental procedures S1.

Nucleic acids extraction

Water samples for DNA were collected in 1L PC bottles (Nalgene) and fixed in a 10% final volume stop-solution (5% of phenol in 99.8% ethanol). The bottles were inverted to ensure complete mixing and stored under dark conditions until filtration (max 24 h). Samples were filtered onto 0.22 µm Durapore GVWP04700 filters (Millipore), which were stored at -80°C. Prior to DNA/RNA

extractions, the filters were cut and vortexed for 10 min with UV treated zirconia beads and RLT Plus Buffer containing β-mercaptoethanol. Total DNA was isolated using the AllPrep DNA/RNA Mini Kit (Qiagen) and on-column DNA digestion (RNase-Free DNase Set, Qiagen) was applied. The second DNA digestion was carried out using the TURBO DNA-free™ kit (Life Technologies) following manufacturer’s instructions. The DNA Clean & Concentrator™ and RNA Clean & Concentrator™ kits (Zymo Research) were used to purify and concentrate total genomic DNA. DNA concentrations were measured using Quant-IT PicoGreen (Invitrogen) and Quant-IT RiboGreen (Invitrogen) assays, respectively.

Sequencing and metagenomic analysis

DNA (2–10 ng) from each sample was prepared with the Rubicon ThruPlex kit (Rubicon Genomics, Ann Arbor, Michigan, USA) according to the instructions of the manufacturer. Cleaning steps were performed with MyOne™ carboxylic acid-coated superparamagnetic beads (Invitrogen, Carlsbad, CA, USA). Libraries were sequenced on a HiSeq 2500 (Illumina Inc., San Diego, CA, USA). On average, 18 million paired-end reads of 2 × 125 bp per sample were generated. Raw reads were quality trimmed with Cutadapt (Martin, 2011) from both read ends and duplicate reads were removed with fastuniq (Xu et al., 2012). High quality reads were mapped onto the BARM database containing the most informative reference genomes in the BS (Alneberg et al., submitted) with Bowtie2 using default parameters (Langmead et al., 2013). The raw

58

counts were calculated from Bedtools histogram output (Quinlan, 2014) and quantitative abundance of reads were annotated using COG protein families (Galperin et al., 2015).

For determining the community composition partial 16S rRNA genes (V3- V4 region) were PCR amplified using the primers non-barcoded Bakt_341F (CCTACGGGNGGCWGCA) and Bakt_805R (GACTACHVGGGTATCTAATCC) (Herlemann et al., 2011). For the barcoding second PCR was carried out using using P5/P7 index (in 96 indicies batches). Products were then sequenced with an Illumina MiSeq paired-end 2 x 250 bp (PE250) multiplex platform in SciLifeLab/NGI (Solna, Sweden).

Raw amplicon reads were quality trimmed with Trimmomatic (ver 0.32). We removed any chimeric reads and assigned the reads into operational taxonomic units (OTUs) using 97% cut-off in cd-hit-otu (Li et al., 2012). We used SINA 1.2 (Pruesse et al., 2012) against the SILVA database (v. 115) to classify unique OTUs and estimated the relative abundance of each OTU using an in-house Python script.

Data analyses

All data and statistical analyses were carried out in R (http://www.r-project.org/). The EdgeR package was used for differential abundance analysis (Robinson et al., 2010) since the data followed a negative binomial distribution. Low abundance COGs were filtered out by only including COGs exceeding 100 counts per million in at least three samples, leaving a total of 2,065 and 2,015 COGs for further analysis for Exp I and Exp II, respectively. Trimmed mean of M-values

(TMM) normalization was performed to eliminate composition biases between libraries. The exact test function was used in order to identify COGs that displayed significantly difference in normalized abundance level between control replicates and treatment triplicates. We extracted differentially abundant COGs at a significance level of p < 0.01, with a false discovery rate (FDR) of <0.05. As an additional quality control step of COGs detected with significantly different abundance for the cyanobacterial lysate treatment, we required that the logFC had to be more than 1.5 compared to the average of controls. In addition, a high number of significant COGs for treatments from Exp II (DOMhum, DOMagri, DOMsea and NP) were identified with initial criteria. In order to shortlist these, only COGs with logFC more than 0.4 compared to the average of controls were extracted.

Cluster analyses were based on normalized abundance profiles of significantly up-represented COGs identified as described above (n = 221 for Exp I and n = 144 for Exp II) and performed using Pearson correlation and hclust function in R. For k-medoids clustering, partitioning around medoid (PAM) algorithm was used with Pearson correlation for distance calculations. The number of clusters was validated using silhouettes (data not shown) (Rousseeuw, 1987).

Transporter examination was performed by extracting transporters from the subset of significantly up-represented COGs (logFC > 0, FDR < 0.05, p-adjusted < 0.001) based on previous indications of their relevance to biogeochemical processes (Poretsky et al.,

59

2010; Gifford et al., 2013; Harke and Gobler, 2013; Smith et al., 2013; Penn et al., 2014; Saito et al., 2014; Satinsky et al., 2014; Shilova et al., 2014; Doxey et al., 2015).

Linear mixed models were fitted to the time series measurements of BA, VA, BP, enzyme activities, amino acid uptake, DOC/DON and inorganic nutrient concentrations using the R package lme4 (Bates et al., 2015). Levels of variables were modelled using the treatment and time as fixed effects and replicates of microcosms as random effects. Significance of fixed effects was assessed by an F-test using a significance level of 5%. Model checking was based on residual plots and normal probability plots using the raw residuals. Models were reduced using the likelihood ratio test. A 1% or 5% significance level was used. Pairwise comparisons were evaluated based on adjusted p-values obtained using the single-step method comparing control treatment with treatments (Table S1). Detrended correspondence analysis (DCA) of COG abundance data and linear fitting of measured abundance/rate/concentration variables was analyzed using vegan package in R. Diversity indices such as abundance-based coverage estimator (ACE) (Chao and Lee, 1992) and Shannon index (Hill, 1973) were calculated and differences between treatments were tested statistically using ANOVA.

Acknowledgements

The authors would like to thank J. Hansen for technical assistance, S. J. Traving for useful discussions, C. Ritz for statistical advice, and the staff of R/V Ophelia and Tvärminne Zoological Station for technical assistance. This work resulted from the BONUS Blueprint

project supported by BONUS (Art 185), funded jointly by the EU and the Danish Council for Independent Research, Estonian Research Council, Swedish Research Council FORMAS, and Academy of Finland. In addition V.K. was supported by a personal grant PUT-134 and 1389 from the Estonian Research Council.

Availability of data and materials

Metagenomic and amplicon sequences were deposited in NCBI SRA (Bioproject number PRJNA435478).

Conflict of interest statement

The authors declare there are no conflicts of interest.

References Alneberg, J,, Sundh, J., Bennke, C., Beier, S.,

Lundin, D., Hugerth, L., Pinhassi, J., Kisand, V., Riemann, L., Jürgens, K., Labrenz, M., and Andersson, A.F (submitted). BARM and BalticMicrobeDB: a reference metagenome and interface to meta-omic data for the Baltic Sea. Scientific Data, in press

Amon, R.M.W. and Benner, R. (1996) Bacterial utilization of different size classes of dissolved organic matter. Limnol. Oceanogr. 41: 41–51.

Anantharaman, K., Brown, C.T., Hug, L.A., Sharon, I., Castelle, C.J., Probst, A.J., et al. (2016) Thousands of microbial genomes shed light on interconnected biogeochemical processes in an aquifer system. Nat. Commun. 7: 13219.

Andersson, A.F., Riemann, L., and Bertilsson, S. (2010) Pyrosequencing reveals contrasting seasonal dynamics of taxa within Baltic Sea bacterioplankton communities. ISME J. 4: 171–181.

Angly, F.E., Willner, D., Prieto-Davó, A., Edwards, R.A., Schmieder, R., Vega-Thurber,

60

R., et al. (2009) The GAAS metagenomic tool and its estimations of viral and microbial average genome size in four major biomes. PLoS Comput. Biol. 5: e1000593.

Bates, D., Mächler, M., Bolker, B., and Walker, S. (2015) Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67: 51.

Bergauer, K., Fernandez-Guerra, A., Garcia, J.A.L., Sprenger, R.R., Stepanauskas, R., Pachiadaki, M.G., et al. (2018) Organic matter processing by microbial communities throughout the Atlantic water column as revealed by metaproteomics. Proc. Natl. Acad. Sci. 115: E400–E408.

Burke, C., Steinberg, P., Rusch, D., Kjelleberg, S., and Thomas, T. (2011) Bacterial community assembly based on functional genes rather than species. PNAS 108: 14288–14293.

Carlsson, P., Segatto, A.Z., and Graneli, E. (1993) Nitrogen bound to humic matter of terrestrial origin - a nitrogen pool for coastal phytoplankton? Mar. Ecol. Prog. Ser. 97: 105–116.

Caruso, G., La Ferla, R., Azzaro, M., Zoppini, A., Marino, G., Petochi, T., et al. (2016) Microbial assemblages for environmental quality assessment: Knowledge, gaps and usefulness in the European Marine Strategy Framework Directive. Crit. Rev. Microbiol. 42: 883–904.

Cases, I., De Lorenzo, V., and Ouzounis, C.A. (2003) Transcription regulation and environmental adaptation in bacteria. Trends Microbiol. 11: 248–253.

Chao, A. and Lee, S.-M. (1992) Estimating the Number of Classes via Sample Coverage. J. Am. Stat. Assoc. 87: 210–217.

Dinsdale, E. a, Edwards, R. a, Hall, D., Angly, F., Breitbart, M., Brulc, J.M., et al. (2008) Functional metagenomic profiling of nine biomes. Nature 452: 629–632.

Doxey, A.C., Kurtz, D.A., Lynch, M.D., Sauder, L.A., and Neufeld, J.D. (2015) Aquatic metagenomes implicate Thaumarchaeota in global cobalamin production. ISME J 9: 461–471.

Dupont, C.L., Larsson, J., Yooseph, S., Ininbergs, K., Goll, J., Asplund-Samuelsson, J., et al. (2014) Functional tradeoffs underpin salinity-driven divergence in microbial community composition. PLoS One 9: e0089549.

Eilers, H., Pernthaler, J., and Amann, R. (2000) Succession of pelagic marine bacteria during enrichment: A close look at cultivation-induced shifts. Appl. Environ. Microbiol. 66: 4634–4640.

Farnelid, H., Bentzon-Tilia, M., Andersson, A.F., Bertilsson, S., Jost, G., Labrenz, M., et al. (2013) Active nitrogen-fixing heterotrophic bacteria at and below the chemocline of the central Baltic Sea. ISME J. 7: 1413–23.

Flamholz, A., Noor, E., Bar-Even, A., Liebermeister, W., and Milo, R. (2013) Glycolytic strategy as a tradeoff between energy yield and protein cost. Proc. Natl. Acad. Sci. U. S. A. 110: 10039–44.

Fortunato, C.S. and Crump, B.C. (2015) Microbial gene abundance and expression patterns across a river to ocean salinity gradient. PLoS One 10: e0140578.

Forward, J.A., Behrendt, M.C., Wyborn, N.R., Cross, R., and Kelly, D.J. (1997) TRAP transporters: A new family of periplasmic solute transport systems encoded by the dctPQM genes of Rhodobacter capsulatus and by homologs in diverse gram-negative bacteria. J. Bacteriol. 179: 5482–5493.

Franzosa, E. a., Hsu, T., Sirota-Madi, A., Shafquat, A., Abu-Ali, G., Morgan, X.C., and Huttenhower, C. (2015) Sequencing and beyond: integrating molecular “omics” for microbial community profiling. Nat. Rev. Microbiol. 13: 360–72.

Fuchs, B.M., Zubkov, M. V., Sahm, K., Burkill, P.H., and Amann, R. (2000) Changes in community composition during dilution cultures of marine bacterioplankton as assessed by flow cytometric and molecular biological techniques. Environ. Microbiol. 2: 191–201.

Fuhrman, J.A., Cram, J.A., and Needham, D.M. (2015) Marine microbial community dynamics

61

and their ecological interpretation. Nat. Rev. Microbiol. 13: 133–146.

Galperin, M.Y., Makarova, K.S., Wolf, Y.I., and Koonin, E. V. (2015) Expanded Microbial genome coverage and improved protein family annotation in the COG database. Nucleic Acids Res. 43: D261–D269.

Gianoulis, T.A., Raes, J., Patel, P. V, Bjornson, R., Korbel, J.O., Letunic, I., et al. (2009) Quantifying environmental adaptation of metabolic pathways in metagenomics. Proc. Natl. Acad. Sci. 106: 1374–1379.

Gifford, S.M., Sharma, S., Booth, M., and Moran, M.A. (2013) Expression patterns reveal niche diversification in a marine microbial assemblage. ISME J. 7: 281–298.

Gilbert, J. a and Dupont, C.L. (2011) Microbial metagenomics: beyond the genome. Ann. Rev. Mar. Sci. 3: 347–371.

Gomez-Consarnau, L., Lindh, M. V, Gasol, J.M., and Pinhassi, J. (2012) Structuring of bacterioplankton communities by specific dissolved organic carbon compounds. Env. Microbiol 14: 2361–2378.

Graham, E.B., Knelman, J.E., Schindlbacher, A., Siciliano, S., Breulmann, M., Yannarell, A., et al. (2016) Microbes as engines of ecosystem function: When does community structure enhance predictions of ecosystem processes? Front. Microbiol. 7: 214. doi: 10.3389/fmicb.2016.00214.

Grote, J., Schott, T., Bruckner, C.G., Glöckner, F.O., Jost, G., Teeling, H., et al. (2011) Genome and physiology of a model Epsilonproteobacterium responsible for sulfide detoxification in marine oxygen depletion zones. Proc. Natl. Acad. Sci. U. S. A. 109: 506–10.

Haaber, J. and Middelboe, M. (2009) Viral lysis of Phaeocystis pouchetii: implications for algal population dynamics and heterotrophic C, N and P cycling. ISME J 3: 430–441.

Haft, D.H., Selengut, J.D., Richter, R.A., Harkins, D., Basu, M.K., and Beck, E. (2013) TIGRFAMs and genome properties in 2013. Nucleic Acids

Res. 41: D387–D395. doi: 10.1093/nar/gks1234.

Harke, M.J. and Gobler, C.J. (2013) Global Transcriptional Responses of the Toxic Cyanobacterium, Microcystis aeruginosa, to Nitrogen Stress, Phosphorus Stress, and Growth on Organic Matter. PLoS One 8: e0069834.

Herlemann, D.P., Labrenz, M., Jürgens, K., Bertilsson, S., Waniek, J.J., and Andersson, A.F. (2011) Transitions in bacterial communities along the 2000km salinity gradient of the Baltic Sea. ISME J. 5: 1571–1579.

Herlemann, D.P.R., Lundin, D., Labrenz, M., Jürgens, K., Zheng, Z., Aspeborg, H., and Andersson, A.F. (2013) Metagenomic De Novo assembly of an aquatic representative of the verrucomicrobial class Spartobacteria. MBio 4: e00569-12.

Herlemann, D.P.R., Manecki, M., Meeske, C., Pollehne, F., Labrenz, M., Schulz-Bull, D., et al. (2014) Uncoupling of bacterial and terrigenous dissolved organic matter dynamics in decomposition experiments. PLoS One 9:.

Hill, M.O. (1973) Diversity and evenness: a unifying notation and its consequences. Ecology 54: 427–432.

Hugerth, L.W., Larsson, J., Alneberg, J., Lindh, M. V, Legrand, C., Pinhassi, J., and Andersson, A.F. (2015) Metagenome-assembled genomes uncover a global brackish microbiome. Genome Biol. 16: 279.

Jiao, N. and Zheng, Q. (2011) The microbial carbon pump: From genes to ecosystems. Appl. Environ. Microbiol. 77: 7439–7444.

Kirchman, D.L., Dittel, A.I., Findlay, S.E.G., and Fischer, D. (2004) Changes in bacterial activity and community structure in response to dissolved organic matter in the Hudson River, New York. Aquat. Microb. Ecol. 35: 243–257.

Kisand, V., Valente, A., Lahm, A., Tanet, G., and Lettieri, T. (2012) Phylogenetic and functional metagenomic profiling for assessing microbial

62

biodiversity in environmental monitoring. PLoS One 7: e0043630.

Labrenz, M., Sintes, E., Toetzke, F., Zumsteg, A., Herndl, G.J., Seidler, M., and Jürgens, K. (2010) Relevance of a crenarchaeotal subcluster related to Candidatus Nitrosopumilus maritimus to ammonia oxidation in the suboxic zone of the central Baltic Sea. ISME J. 4: 1496–1508.

Landa, M., Cottrell, M.T., Kirchman, D.L., Kaiser, K., Medeiros, P.M., Tremblay, L., et al. (2014) Phylogenetic and structural response of heterotrophic bacteria to dissolved organic matter of different chemical composition in a continuous culture study. Environ. Microbiol. 16: 1668–1681.

Langmead, B., Salzberg, S.L., and Langmead (2013) Bowtie2. Nat. Methods 9: 357–359.

Lauro, F.M., McDougald, D., Thomas, T., Williams, T.J., Egan, S., Rice, S., et al. (2009) The genomic basis of trophic strategy in marine bacteria. Proc. Natl. Acad. Sci. U. S. A. 106: 15527–33.

Li, W., Fu, L., Niu, B., Wu, S., and Wooley, J. (2012) Ultrafast clustering algorithms for metagenomic sequence analysis. Brief. Bioinform. 13: 656–668.

Lindh, M. V., Figueroa, D., Sjöstedt, J., Baltar, F., Lundin, D., Andersson, A., et al. (2015) Transplant experiments uncover Baltic Sea basin-specific responses in bacterioplankton community composition and metabolic activities. Front. Microbiol. 6: 223. doi: 10.3389/fmicb.2015.00223.

Martin, M. (2011) Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17: 10.

McCarren, J., Becker, J.W., Repeta, D.J., Shi, Y., Young, C.R., Malmstrom, R.R., et al. (2010) Microbial community transcriptomes reveal microbes and metabolic pathways associated with dissolved organic matter turnover in the sea. Proc. Natl. Acad. Sci. U. S. A. 107: 16420–7.

Middelboe, M. and Jørgensen, N.O.G. (2006) Viral lysis of bacteria: an important source of dissolved amino acids and cell wall

compounds. J. Mar. Biol. Assoc. UK 86: 605–612.

Morris, R.M., Nunn, B.L., Frazar, C., Goodlett, D.R., Ting, Y.S., and Rocap, G. (2010) Comparative metaproteomics reveals ocean-scale shifts in microbial nutrient utilization and energy transduction. ISME J. 4: 673–685.

Mulligan, C., Fischer, M., and Thomas, G.H. (2011) Tripartite ATP-independent periplasmic (TRAP) transporters in bacteria and archaea. FEMS Microbiol. Rev. 35: 68–86.

Penn, K., Wang, J., Fernando, S.C., and Thompson, J.R. (2014) Secondary metabolite gene expression and interplay of bacterial functions in a tropical freshwater cyanobacterial bloom. ISME J. 8: 1866–1878.

Pinhassi, J. and Berman, T. (2003) Differential growth response of colony-forming alpha- and gamma-proteobacteria in dilution culture and nutrient addition experiments from Lake Kinneret (Israel), the eastern Mediterranean Sea, and the Gulf of Eilat. Appl. Environ. Microbiol. 69: 199–211.

Poretsky, R.S., Sun, S., Mou, X., and Moran, M.A. (2010) Transporter genes expressed by coastal bacterioplankton in response to dissolved organic carbon. Environ. Microbiol. 12: 616–627.

Pruesse, E., Peplies, J., and Glöckner, F.O. (2012) SINA: Accurate high-throughput multiple sequence alignment of ribosomal RNA genes. Bioinformatics 28: 1823–1829.

Quinlan, A.R. (2014) BEDTools: The Swiss-Army tool for genome feature analysis. Curr. Protoc. Bioinforma. 2014: 11.12.1-11.12.34.

Raes, J. and Bork, P. (2008) Molecular eco-systems biology: towards an understanding of community function. Nat. Rev. Microbiol. 6: 693–699.

Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010) edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26: 139–40.

Rousseeuw, P.J. (1987) Silhouettes: A graphical aid to the interpretation and validation of

63

cluster analysis. J. Comput. Appl. Math. 20: 53–65.

Saito, M.A., McIlvin, M.R., Moran, D.M., Goepfert, T.J., DiTullio, G.R., Post, A.F., and Lamborg, C.H. (2014) Multiple nutrient stresses at intersecting Pacific Ocean biomes detected by protein biomarkers. Science 345: 1173–7.

Satinsky, B.M., Crump, B.C., Smith, C.B., Sharma, S., Zielinski, B.L., Doherty, M., et al. (2014) Microspatial gene expression patterns in the Amazon River Plume. Proc. Natl. Acad. Sci. 111: 11085–90.

Shilova, I.N., Robidart, J.C., James Tripp, H., Turk-Kubo, K., Wawrik, B., Post, A.F., et al. (2014) A microarray for assessing transcription from pelagic marine microbial taxa. ISME J. 8: 1476–1491.

Sjöstedt, J., Koch-Schmidt, P., Pontarp, M., Canbäck, B., Tunlid, A., Lundberg, P., et al. (2012) Recruitment of members from the rare biosphere of marine bacterioplankton communities after an environmental disturbance. Appl. Environ. Microbiol. 78: 1361–1369.

Smith, D.P., Thrash, J.C., Nicora, C.D., Lipton, M.S., Burnum-Johnson, K.E., Carini, P., et al. (2013) Proteomic and transcriptomic analyses of “Candidatus Pelagibacter ubique” describe the first PII-independent response to nitrogen limitation in a free-living alphaproteobacterium. MBio 4: e00133-12.

Teeling, H. et al. (2012) Substrate-Controlled Succession of Marine Bacterioplankton Populations Induced by a Phytoplankton Bloom. Science (80-. ). 336: 608–611.

Tringe, S.G., von Mering, C., Kobayashi, A., Salamov, A.A., Chen, K., Chang, H.W., et al. (2005) Comparative metagenomics of microbial communities. Science (80-. ). 308: 554–557.

Won, N. Il, Kim, K.H., Kang, J.H., Park, S.R., and Lee, H.J. (2017) Exploring the impacts of anthropogenic disturbance on seawater and sediment microbial communities in Korean coastal waters using metagenomics analysis.

Int. J. Environ. Res. Public Health 14: 130. doi: 10.3390/ijerph14020130.

Xu, H., Luo, X., Qian, J., Pang, X., Song, J., Qian, G., et al. (2012) FastUniq: A Fast De Novo Duplicates Removal Tool for Paired Short Reads. PLoS One 7: e0052249.

64

Supporting Information

65

66

Supporting information

Coupling biogeochemical process rates and metagenomic blueprints of coastal bacterial assemblages in the context of environmental change Trine Markussen1, Elisabeth M. Happel1, Jonna E. Teikari2, Vimala Huchaiah3, Johannes Alneberg4, Anders F. Andersson4, Kaarina Sivonen2, Lasse Riemann1, Mathias Middelboe1*, Veljo Kisand3* *Correspondence: Mathias Middelboe: mmiddelboe@bio.ku.dk, Veljo Kisand: kisand@ut.ee Table S1. In situ and T0 abundance and activity of bacterioplankton and viruses, concentrations of nutrients and environmental conditions. Mean values ± standard deviation for triplicate samples are shown.

Table S2. Dunnett’s test.

Exp I Exp II

Estimate Std. Error z value Pr(>|z|) Estimate Std. Error z value Pr(>|z|)

AA

uptake

DOMagri 11.88083 2.89235 4.108 0.00129 7.1989 1.9573 3.678 0.00731 DOMhum 9.69551 2.89235 3.352 0.0232 3.5856 1.9573 1.832 0.76638 DOMsea 1.31483 2.89235 0.455 1 5.0925 1.9573 2.602 0.19978 LowO2 2.8227 2.89235 0.976 0.99995 -5.6429 1.9573 -2.883 0.09705 LTC 26.2347 2.89235 9.07 < 0,001 7.2966 1.9573 3.728 0.00603 NP 6.55959 2.89235 2.268 0.40536 3.425 1.9573 1.75 0.82754

BP

DOMagri 0.30204 0.16344 1.848 0.868 0.8158 0.21235 3,842 <0,01 DOMhum -0.3519 0.16344 -2.153 0.624 0.72463 0.21235 3,412 0.0314 DOMsea 0.13715 0.16344 0.839 1 0.65294 0.21235 3,075 0.0892 LowO2 0.19657 0.16344 1.203 1 -0.16964 0.21235 -7.99 1

Øresund (Exp I) Storfjärden (Exp II)

In situ Control DOMhum DOMagri DOMsea LTC NP LowO2

In situ Control DOMhum DOMagri DOMsea LTC NP LowO2

Bacterial abundance (× 10^6 cells mL-1)

1.80 ± 0.13

0.47 ± 0.09 0.52 ± 0.10

0.54 ± 0.11

0.49 ± 0.10

0.21 ± 0.05

0.59 ± 0.03

0.40 ± 0.01

3.3 ± 0.26

0.72 ± 0.07 0.77 ± 0.08

0.73 ± 0.13

0.85 ± 0.13

0.90 ± 0.04

0.72 ± 0.03

0.85 ± 0.05

Bacterial production (µg C L-1 d-1)

1.95 ± 0.06

0.95 ± 0.32 1.0 ± 0.35

1.18 ± 0.38

1.0 ± 0.33

0.80 ± 0.55

1.21 ± 0.29

0.89 ± 0.11

4.1 ± 0.3

1.41 ± 0.14 1.41 ± 0.12

1.20 ± 0.02

1.23 ± 0.05

0.45 ± 0.24

1.51 ± 0.26

1.79 ± 0.39

Viral abundance (× 10^6 viruses mL-1)

13.13 ± 1.0

12.43 ± 2.58 19.3 ± 2.16

16.0 ± 2.98

24.63 ± 5.31

14.65 ± 1.06

8.36 ± 0.3

9.49 ± 1.31

15.5 ± 2.0

12.25 ± 0.63

11.91 ± 1.44

29.85 ± 4.40

14.27 ± 0.65

15.23 ± 0.21

13.14 ± 0.57

13.30 ± 0.21

DOC (µM) 381 ± 7 382 ± 14 553 ± 51 482 ± 78 365 ± 4

552 ± 14

377 ± 2

374 ± 12

667 ± 91

556 ± 81 711 ± 67 756 ± 59 610 ± 28

692 ± 7

624 ± 19

622 ± 12

DON (µM) 50 ± 2 47 ± 0.4 52 ± 1 104 ± 16 47 ± 0.5

68 ± 1

56 ± 1 47 ± 1 50 ± 2 48 ± 2 52 ± 1 57 ± 1 49 ± 1

58 ± 0.5

58 ± 1 48 ± 2

NH 4 + (µM)

0.24 ± 0.01

0.57 ± 0.03 0.94 ± 0.13

1.03 ± 0.04

0.58 ± 0.01

0.70 ± 0.04

1.58 ± 0.02

0.62 ± 0.04

0.26 ± 0.01

0.36 ± 0.03 0.54 ± 0.03

1.12 ± 0.01

0.34 ± 0.02

0.36 ± 0.01

1.30 ± 0.02

0.33 ± 0.01

NO3 - (µM) 2.1 ± 0.2

1.74 ± 0.19 3.63 ± 0.66

61.87 ± 9.35

2.05 ± 0.10

2.45 ± 0.97

4.73 ± 0.74

1.67 ± 0.06

0.44 ± 0.03

0.27 ± 0.08 2.61 ± 0.66

0.56 ± 0.08

0.27 ± 0.16

0.21 ± 0.13

6.19 ± 2.90

0.30 ± 0.10

PO4 3- (µM)

0.09 ± 0.02

0.06 ± 0.11 0.20 ± 0.03

0.19 ± 0.08

0.21 ± 0.06

0.63 ± 0.06

0.14 ± 0.04

0.14 ± 0.11

0.26 ± 0.03

0.28 ± 0.03 0.26 ± 0.04

0.26 ± 0.03

0.18 ± 0.01

0.66 ± 0.02

0.26 ± 0.09

0.20 ± 0.05

67

LTC 2.18627 0.16344 13.377 <0,01 2.39307 0.21235 11,270 <0,01 NP 0.20665 0.16344 1.264 1 0.31616 0.21235 1,489 0.9948

VA

DOMagri 0.717791 0.114566 6.265 <0,01 0.859501 0.062513 13.749 <0,01 DOMhum 0.386198 0.114566 3.371 0.0317 0.089466 0.062513 1.431 0.9977 DOMsea 0.471313 0.114566 4.114 <0,01 0.125231 0.062513 2.003 0.7859 LowO2 0.049961 0.114566 0.436 1 0.076527 0.062513 1.224 0.9999 LTC 1.459634 0.114566 12.741 <0,01 0.188378 0.072183 2.61 0.2904 NP 0.300773 0.114566 2.625 0.2523 0.025993 0.062513 0.416 1

BA

DOMagri 0.376623 0.113108 3.33 0.03663 0.68146 0.15504 4.395 <0,01 DOMhum 0.190873 0.113108 1.688 0.94311 0.4642 0.15504 2.994 0.113 DOMsea 0.324981 0.113108 2.873 0.13853 0.86932 0.15504 5.607 <0,01 LowO2 -0.053192 0.113108 -0.47 1 0.12032 0.15504 0.776 1 LTC 1.301668 0.113108 11.508 < 0,001 4.85517 0.1791 27.109 <0,01 NP 0.133649 0.113108 1.182 0.99992 0.26977 0.15504 1.74 0.9451

DO

C

DOMagri 0.53085 0.1359 3.906 0.00184 0.27885 0.13652 2.043 0.44745 DOMhum 0.52921 0.1359 3.894 0.00185 0.27856 0.13652 2.04 0.44902 DOMsea 0.10436 0.1359 0.768 0.99992 -0.22916 0.13652 -1.679 0.74239 LowO2 0.10182 0.1359 0.749 0.99994 -0.04581 0.13652 -0.336 1 LTC 0.16325 0.1359 1.201 0.97804 -0.74522 0.13652 -5.459 < 0,001 NP 0.07946 0.1359 0.585 1 -0.5063 0.13652 -3.709 0.00388

DO

N

DOMagri 0.95441 0.06422 14,860 <0,001 0.15306 0.04507 3.396 0.0123 DOMhum 0.15627 0.06422 2,433 0.204 0.05984 0.04507 1.328 0.9458 DOMsea 0.05382 0.06422 0.838 0.999 -0.05095 0.04507 -1.13 0.9881 LowO2 0.03995 0.06422 0.622 1,000 0.01081 0.04507 0.24 1 LTC 0.29962 0.06422 4,665 <0,001 -0.13226 0.04507 -2.935 0.0549 NP 0.16807 0.06422 2,617 0.131 -0.06637 0.04507 -1.473 0.8813

NH

4

DOMagri -0.08246 0.11806 -0.69 1 0.65883 0.12331 5.343 < 0,001 DOMhum 0.43635 0.11806 3.69 0.0067 0.01549 0.12331 0.126 1 DOMsea 0.13984 0.11806 1.18 0.9981 0.13703 0.12331 1.111 0.99932 LowO2 -0.20523 0.11806 -1.73 0.8348 0.38451 0.12331 3.118 0.04901 LTC 0.02475 0.11806 0.21 1 1.85517 0.12331 15.044 < 0,001 NP 0.52879 0.11806 4.47 < 0,001 1.35054 0.12331 10.952 < 0,001

PO4

DOMagri 0.34065 0.48682 0.7 1 0.05388 0.18729 0.288 1 DOMhum 0.11204 0.48682 0.23 1 -0.00611 0.18729 -0.033 1 DOMsea 1.06035 0.48682 2.178 0.48366 -0.05151 0.18729 -0.275 1 LowO2 0.19409 0.48682 0.399 1 0.02206 0.18729 0.118 1 LTC 1.5395 0.48682 3.162 0.04281 1.18153 0.18729 6.308 <0,001 NP 0.13608 0.48682 0.28 1 0.21132 0.18729 1.128 0.9992

NO

3

DOMagri 2.85861 0.09626 29.698 < 0,001 0.07307 0.32405 0.225 1 DOMhum 0.18415 0.09626 1.913 0.698 1.22453 0.32405 3.779 0.00488 DOMsea -0.04892 0.09626 -0.508 1 -1.59476 0.32405 -4.921 < 0,001 LowO2 0.03983 0.09626 0.414 1 -0.90209 0.32405 -2.784 0.12529 LTC -0.05811 0.09626 -0.604 1 -1.30425 0.32405 -4.025 0.00188 NP 0.78477 0.09626 8.153 < 0,001 2.43719 0.32405 7.521 < 0,001

Protease

DOMagri 1.8698 0.4125 4.533 < 0,001 4.2984 1.0006 4.296 < 0,001 DOMhum -0.2465 0.4125 -0.597 1 -0.5606 1.0006 -0.56 1 DOMsea 3.7909 0.4125 9.191 < 0,001 3.1416 1.0006 3.14 0.04609 LowO2 1.6092 0.4125 3.901 0.00316 1.1359 1.0006 1.135 0.9991 LTC 10.3127 0.4125 25.002 < 0,001 14.5957 1.0006 14.587 < 0,001 NP 0.3831 0.4125 0.929 0.99998 0.8946 1.0006 0.894 0.99999

Beta-glucosidase

DOMagri 0.15447 0.12937 1.194 0.9974 0.81813 0.13919 5.878 < 0,001 DOMhum 0.30263 0.12937 2.339 0.3415 0.37351 0.13919 2.683 0.16206 DOMsea 0.19897 0.12937 1.538 0.9344 0.37927 0.13919 2.725 0.14627 LowO2 -0.16306 0.12937 -1.26 0.994 0.04523 0.13919 0.325 1 LTC 1.99362 0.12937 15.41 <0,001 7.83846 0.13919 56.314 < 0,001 NP 0.07135 0.12937 0.551 1 0.36465 0.13919 2.62 0.18908

Alkaline

phosphatase

DOMagri 0.92215 0.17004 5.423 < 0,001 -2.5843 0.2778 -9.302 <0,001 DOMhum 0.10136 0.17004 0.596 1 -2.472 0.2778 -8.897 <0,001 DOMsea 1.82659 0.17004 10.742 < 0,001 -5.5291 0.2778 -19.9 <0,001 LowO2 0.81024 0.17004 4.765 < 0,001 -0.6028 0.2778 -2.17 0.483 LTC 6.0769 0.17004 35.739 < 0,001 0.2199 0.2778 0.791 1 NP 0.01935 0.17004 0.114 1 -1.8878 0.2778 -6.795 <0,001

68

Table S3. COGs that were statistically (p < 0.01; FDR > 0.05) up-represented in experimental treatments when compared to the control microcosms. COGs are listed after additional quality control step (details in Experimental procedures). LogFC are listed relative to controls and k-medoids cluster are indicated as number 1-5 (see Fig. 3). This table can be provided upon request Table S4. Up-represented DOC-related COGs and transporters. This table can be provided upon request Table S5. Association between measured bulk variables/rates and abundance of significantly up- or down-represented COGs. COGs were clustered using DCA analysis (see also Fig. 5) and bulk variables/rates fitted linearly with COG scores. This table can be provided upon request

Figure S1. Map of the sampling and experiment sites. DOMhum site - Lappväärti river in Finland used to collect pristine humic rich river water, DOMagri site – Leilupe river in Latvia influenced by agricultural pollution. Exp I (Helsingør) experiment site with Øresund microbial community, Exp II (Tvärminne) experiment site with microbial community from Storfjärden, Gulf of Finland.

69

Figure S2. Schematic illustration of the experimental setup performed at two sites in the Baltic Sea. Treatments were performed in triplicates and controls as six replicates. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous dissolved organic matter; LTC, cell lysate from cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration; Control, controls. A detailed description of the experiment is provided in Experimental procedures.

Figure S3. Total DOC consumption in Exp I (left) and Exp II (right). The DOC consumption was calculated from O2 consumption based on oxygen measurements (details in Experimental procedures S1).

70

Figure S4. Inorganic nutrient concentrations in microcosms from Exp I (left panel) and Exp II (right panel). Ammonium concentration is shown in upper panel, nitrate concentration in the middle panel and phosphate concentration in the bottom panel. Details of significant differences between treatments and controls (Dunnett’s test) can be seen in Table S1. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous dissolved organic matter; LTC, cell lysate from cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration; Control, controls.

71

Figure S5. Taxonomic richness and diversity comparison. A) Abundance-based coverage estimator (ACE) of bacterial communities in treatments from Exp I (red) and Exp II (blue) of triplicate samples. B) Shannon index of bacterial communities in treatments from Exp I (red) and Exp II (blue) of triplicate samples. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous dissolved organic matter; LTC, cell lysate from cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration

72

Figure S6. MDS plot of treatments from Exp I (circles) and Exp II (triangles) based on COG functional profiles. Distances correspond to leading log2-fold-changes (logFC) between each sample. Controls (black), DOMagri (yellow), DOMhum (red), DOMsea (orange), LowO2 (purple), LTC (green), NP (blue).

Figure S7. COGs assigned to selected major functional categories as a percent of total number of significant COGs displaying higher abundance compared to controls in a) Exp I and b) Exp II. The ‘Transporter and metabolism’ category sums percent COGs related to Amino acid transport and metabolism (E), Carbohydrate transport and metabolism (G), Coenzyme transport and metabolism (H), Energy production and conversion (C), Inorganic ion transport and metabolism (P), Lipid transport and metabolism (I), Nucleotide transport and metabolism (F), Secondary metabolites biosynthesis, transport and catabolism (Q). The ‘Others’ category sums percent COGs related to Cytoskeleton (Z), Extracellular structures (W), Chromatin structure and dynamics (B), RNA processing and medication (A), Posttranslational modification, protein turnover, chaperones (O) and not annotated (NA). Data represent average from triplicate microcosms.

73

Figure S8. Clustering of COGs significantly up-represented in one or more treatments compared to controls for (A) Exp I (n = 221) and (B) Exp II (n = 144). The COGs have been clustered according to treatment into five clusters by k-medoids clustering for Exp I (E) and Exp II (F). The y axis displays COG abundances and the x axis shows the different treatments. DOMhum, fresh water from humic rich river; DOMagri, fresh water from polluted river influenced by agriculture; DOMsea, autochthonous dissolved organic matter; LTC, cell lysate from cyanobacterial culture; NP, inorganic nutrient addition; LowO2, reduced O2 concentration; Control, controls.

74

Figure S9. Venn diagrams of overlapping significant COGs between treatments from Exp I and Exp II either up- or down-represented relative to controls. The overlapping COGs for DOMhum were COG0235, COG1638, COG3772, COG1346, COG1014, COG4231, COG2828, COG5410, COG2358, COG4666, COG4626, COG3740, COG4695, COG5545, COG1783, COG5511, COG5525, COG5323 and COG0234 (up-represented) and COG1494 and COG1376 (down-represented. The overlapping COGs for DOMagri were COG2195 (down-represented) and COG0222 (up-represented). For DOMsea overlapping COGs were COG2849, COG2195, COG1629, COG5520, COG3537, COG2356 (up-represented) and COG1690 (down-represented). LTC showed a huge overlap and an additional filter was applied resulting in 19 overlapping COGs with 9 up represented (COG2819, COG3508, COG4152, COG2849, COG4704, COG4874, COG3757, COG1368 and COG1979) and 10 down-represented (COG1180, COG2401, COG5126, COG1276, COG2519, COG2368, COG0638, COG1405, COG5059 and COG5534). The overlapping COGs of NP treatment were COG3772 and COG1783 (up-represented) and no overlapping COG down represented. No overlapping COG was identified for treatment LowO2. See Table S2 for annotations.

75

Experimental procedures S1

Bacterial and viral abundance

Bacteria and viruses were counted by flow cytometry (Marie et al., 1999). Samples were fixed with 1% glutaraldehyde (final concentration, Sigma-Aldrich) and stored at -80°C until analysis. For bacterial counts, thawed samples were diluted 10 to 100-fold in autoclaved 0.2 µm filtered 1X TE buffer (Sigma-Aldrich) and stained with SYBR Green (Invitrogen-Molecular probes) for 10 min. Samples for viral counts were heated to 80°C for 10 min before the stained samples were analysed using a BD FACSCanto II flow cytometer with the flow rate setting “low”. The trigger was set on the green fluorescence, and samples were analysed for 1 min. The actual sample flow rate was measured using BD Trucount™ Absolute Counting tubes (BD Biosciences) to be 9 ± 1.3 µL min-1. The blank consisted of 0.2 µm filtered 1X TE-buffer and SYBR-Green and was analysed the same way as the samples.

Bacterial production

Bacterial production was measured by [3H]-thymidine incorporation (Fuhrman and Azam, 1982), as modified for microcentrifugation (Smith and Azam, 1992). Triplicates of 1.2 mL aliquots of samples were incubated with 10 nM [3H]-thymidine (final concentration, PerkinElmer) for 1 hour at in situ temperature in darkness. The reaction was stopped by addition of trichloracetic acid (TCA) to a final concentration of 5%. Samples with 5% TCA added prior to the addition of [3H]-thymidine served as blanks. Carbon production was calculated as described by Simon and Azam (Simon and Azam, 1989)

using 1.1 × 1018 cells µmol-1 thymidine (Riemann et al., 1987) and 20 fg C bacterium-1 (Lee and Fuhrman, 1987).

Extracellular enzymatic activity

The potential activities of β-glucosidase, leucine aminopeptidase, and alkaline phosphatase) were measured using the fluorogenic substrate analogues 4-methyl-umbelliferyl-β-D-glucopyranoside (MUD), L-leucine-4-methylcoumarinylamid hydrochloride (Leu-MCA), and 4-methyl-umbelliferyl phosphate (4-MUD), respectively. The enzyme assays were performed in four replicates with blanks in duplicates for each enzyme and were run in black 96-well microplates. Samples were incubated in the dark at in situ temperature with either 100 µM MUD, 200 µM 4-MUD, or 200 µM Leu-MCA (final concentration; experimentally determined to be saturating in the studied waters) for 3-5 h.

Oxygen saturation

Oxygen saturation was measured in triplicate subsamples from each batch experiment incubated in gas tight 20 mL glass tubes in a water bath at in situ temperature to keep temperature constant and to minimize oxygen contamination during incubation. The measurements were made using PyroScience FireStingO2 oxygen sensors calibrated with a 100% saturated O2 sample (air bubbled sample water) and a 0% saturation sample (sample water saturated with Na2SO3) and the program Pyro Oxygen Logger Software version 3.07 (Pyroscience). Oxygen measurements were normalized to a 100% saturation value recorded for each bottle

76

before starting the experiment. Total DOC consumption was calculated from O2 consumption based on above measurements after 65 days (Exp I) and 42 days (Exp II) of incubation.

Amino acid uptake

Amino acid uptake was estimated by incubation of subsamples with a mixture of carbon-14 labelled L-amino acids (50 µCi; PerkinElmer). 9 nCi was added to 10 mL subsample and incubated for 30-45 min at in situ temperature in darkness. Incubations were stopped with glutaraldehyde (0.2% final concentration) and filtered through a 0.2 µm mixed cellulose ester filter (Advantec) at low vacuum and rinsed four times with Milli-Q water. The filters were added to 5 mL scintillation cocktail and counted after 24 h. Amino acid uptake rates (% h-1, not shown) were used in the detrended correspondence analysis.

Chemical analyses

Ammonium (NH4+) concentrations were measured in fresh samples using ortho-phthaladehyde (Holmes et al., 1999) and a rapid flow analyser (Turner Designs Trilogy® Laboratory Fluorometer). Nitrate (NO3-) and phosphate (PO43-) was measured on an autoanalyser following published procedures for NO3- (Wood et al., 1967) and PO43- (Murphy and Riley, 1962). Dissolved organic carbon (DOC) and dissolved organic nitrogen (DON) were measured on a Shimadzu TOC-L Total Organic Carbon Analyzer (Shimadzu Corporation) as previous described (Paulsen et al., 2017). Combined intra- and extracellular concentrations of nodularin were determined by extracting toxins from freeze-dried samples to 70 % methanol and analyzed by UPLC system (a more detailed description of toxin

analysis are presented in Experimental procedures S1).

Toxin analysis

Combined intra- and extracellular concentrations of nodularin were determined by extracting toxins from freeze-dried samples to 70 % methanol at 80 °C for 1 hour. Samples were injected into Acquity UPLC system (Waters, Manchester, UK), equipped with Kinetex® 1.7 µm C8 100 Å, 50 x 2.1 mm LC Column. The UPLC was operated with a flow-rate of 0.3 ml/min in gradient mode, at a temperature of 40 °C. Solvents used in the gradient were A: 0.1 % formic acid in water and B: 0.1 % formic acid in 1 to 1 mixture of acetonitrile and isopropanol. The initial conditions of the linear gradient were A: 25 % and B: 75 % and the conditions were changed to A: 35 % and B: 65 % in 5 min. Injection volume was 1 μL. Mass spectra were recorded with Waters SynaptG2- Si mass spectrometer (Waters, Manchester, UK). Measurements were performed using negative electrospray ionization (ESI) in resolution mode and ions were scanned in the range from 500 to 1300 m/z. MS analyses were performed with scan time of 0.1 s. Capillary voltage was 2.0kV, source temperature 120 °C, sampling cone 40.0, source offset 80.0, desolvation temperature 600 °C, desolvation gas flow 1000 l/h and nebulizer gas flow 6.5 Bar. Leucine-encephalin was used as a lock mass and calibration was done with mixture of sodium formiate and ultramark® 1621. Standard curve containing dilution series of known nodularin concentrations was run alongside toxin samples. All the samples, including standards, were spiked by nostophycin to ensure successful extraction and device functioning.

77

References

Fuhrman, J.A. and Azam, F. (1982) Thymidine Incorporation as a Measure of Heterotrophic Bacterioplankton Production in Marine Surface Waters - Evaluation and Field Results. Mar. Biol. 66: 109–120.

Holmes, R.M., Aminot, A., Kerouel, R., Hooker, B.A., and Peterson, B.J. (1999) A simple and precise method for measuring ammonium in marine and freshwater ecosystems. Can. J. Fish. Aquat. Sci. 56: 1801–1808.

Lee, S. and Fuhrman, J.A. (1987) Relationships between Biovolume and Biomass of Naturally Derived Marine Bacterioplankton. Appl Env. Microbiol 53: 1298–1303.

Marie, D., Brussaard, C.P.D., Thyrhaug, R., Bratbak, G., and Vaulot, D. (1999) Enumeration of marine viruses in culture and natural samples by flow cytometry. Appl. Environ. Microbiol. 65: 45–52.

Murphy, J. and Riley, J.P. (1962) A modified single solution method for the determination of phosphate in natural waters. Anal. Chim. Acta 27: 31–36.

Paulsen, M.L., Nielsen, S.E.B., Müller, O., Møller, E.F., Stedmon, C.A., Juul-Pedersen, T., et al. (2017) Carbon Bioavailability in a High Arctic Fjord Influenced by Glacial Meltwater, NE Greenland. Front. Mar. Sci. 4: 176.

Riemann, B., Bjørnsen, P.K., Newell, S., and Fallon, R. (1987) Calculation of cell production of coastal marine bacteria based on measured incorporation of [3H]thymidine. Limnol. Oceanogr. 32: 471–476.

Simon, M. and Azam, F. (1989) Protein-Content and Protein-Synthesis Rates of Planktonic Marine-Bacteria. Mar. Ecol. Prog. Ser. 51: 201–213.

Smith, D.C. and Azam, F. (1992) A simple, economical method for measuring bacterial protein synthesis rates in seawater using 3H-leucine. Mar. Microb. Food Webs 6: 107–114.

Wood, E.D., Armstrong, F.A.J., and Richards, F.A. (1967) Determination of Nitrate in Sea Water by Cadmium-Copper Reduction to Nitrite. J.

Mar. Biol. Assoc. UK 47: 23–31.

78

Paper II

79

80

Effects of allochthonous DOM input on microbial composition and nitrogen cycling genes at two contrasting estuarine sites Elisabeth M. Happel1, Trine Markussen1, Jonna E. Teikari2, Vimala Huchaiah3, Johannes Alneberg4, Anders F. Andersson4, Kaarina Sivonen2, Matthias Middelboe1, Veljo Kisand3, Lasse Riemann1*.

1Marine Biology Section, Department of Biology, University of Copenhagen, Helsingør, Denmark 2University of Helsinki, Division of Microbiology, Helsinki, Finland 3University of Tartu, Institute of Technology, Tartu, Estonia 4KTH Royal Institute of Technology, Science for Life Laboratory, Department of Gene Technology, School of Engineering Sciences in Chemistry, Biotechnology and Health, Stockholm, Sweden

*Correspondence: Lasse Riemann, lriemann@bio.ku.dk Heterotrophic bacteria are important drivers of nitrogen (N) cycling and the processing of dissolved organic matter (DOM). Projected increases in precipitation will potentially cause increased loads of riverine DOM to the Baltic Sea and likely affect the composition and function of bacterioplankton communities. To investigate this, the effects of riverine DOM from two different catchment areas (agricultural and forest) on natural bacterioplankton assemblages from two contrasting sites in the Baltic Sea were examined. Two microcosm experiments were carried out, where the community composition (16S rRNA gene sequencing), the composition of a suite of N cycling genes (metagenomics), and the abundance and expression of amoA genes (quantitative PCR) were investigated. The river water treatments evoked a significant response in bacterial growth, but effect on overall community composition and on the representation of a suite of N cycling genes were insignificant. Instead, treatment effects were reflected in the prevalence of specific taxonomic families, specific N related functions, and in the expression of amoA genes. The study suggests that bacterioplankton responses to changes in the DOM pool are constrained to part of the bacterial community, whereas most taxa remain relatively unaffected.

Introduction

Marine heterotrophic bacterioplankton process dissolved organic matter (DOM), thereby mineralizing nutrients essential for growth of phytoplankton and affecting overall productivity in marine waters (Azam et al., 1983). Among nutrients, nitrogen (N) is

a primary constituent of various cellular macromolecules, and the availability of N is commonly a limiting factor for primary and secondary production in diverse marine systems (Ryther and Dunstan, 1971; Bristow et al., 2017). In marine coastal systems, the release of dissolved inorganic N (DIN)

81

through the degradation of dissolved organic nitrogen (DON) can be orders of magnitude higher than the input of DIN from land (Knudsen-Leerbeck et al., 2017). Hence, N released or acquired through microbial degradation of nitrogenous DOM is an important N source for bacterioplankton and phytoplankton growth (Bronk et al., 2007). Bacterioplankton control not only the accessibility of N through DOM processing, but also regulate the oxidative state of N present in the environment through a series of oxidative and reductive processes. Consortia of microorganisms thereby mediate key steps in the marine nitrogen cycle (Falkowski et al., 2008), including e.g. ammonification, biological nitrogen fixation (BNF), nitrification, and denitrification (Zehr and Ward, 2002), and ultimately determine the availability of N for higher trophic levels, e.g. phytoplankton. For instance, the form and oxidation level, e.g. whether inorganic N is available as ammonia or nitrate, may affect both the productivity and the composition of the phytoplankton community (Glibert et al., 2016).

In estuarine environments, riverine DOM is an important source of highly labile N (Seitzinger et al., 2005; Bronk et al., 2007). The characteristics of the riverine DOM may depend on the catchment area and on season. Consequently, it is conceivable that the riverine input, particularly in N limited environments, selects for bacterioplankton capable of hydrolyzing nitrogenous DOM and for taxa involved in down-stream nitrogen cycling processes. Moreover, the bacterial community response would likely rely on the availability and nature of the nitrogenous DOM and depend on bacterioplankton community composition and contemporary environmental conditions.

Hence, responses are expected to differ between localities. While these assumptions appear logical, they have to our knowledge not been experimentally verified.

The Baltic Sea is the second largest estuarine system in the world and encompasses separate sub-basins with unique geology and a strong north-south salinity gradient driven by river outlets (Rönnberg and Bonsdorff, 2004). The north is characterized by high DOM concentrations and phosphorous (P) limited plankton production whereas the south has lower DOM concentrations and N limited plankton productivity (Bernes, 2005; Rowe et al., 2018). Further, catchment characteristics vary from primarily forest in the north to agricultural landscapes in the south. The gradient in biogeochemistry is also reflected in extensive changes in microbial community composition from north to south (Herlemann et al., 2011). Climate change is predicted to increase precipitation and the allochthonous DOM input via rivers to the Baltic Sea (Andersson et al., 2015). The loading and characteristics of the DOM will likely affect the future microbial community composition, activity, DOM utilization, and nutrient biogeochemistry in the Baltic Sea (Traving et al., 2017; Rowe et al., 2018). However, responses will likely vary between north and south due to differences in catchments, characteristics of the incoming DOM, and composition of the recipient microbial communities.

In the present study we examined effects of riverine DOM loading in incubation experiments in two contrasting environments; the southern Baltic Sea (Øresund) after the spring bloom and in the northern Baltic Sea (Storfjärden) in summer

82

(Fig. S1). After the spring bloom, Øresund surface water is typically N depleted whereas the surface water at Storfjärden typically has higher N and DOM concentrations during the summer. We tested the effects of allochthonous DOM loads at both sites by additions of river water from an agricultural and a forest (humic) catchment area, and examined the microbial community composition (16S rRNA genes), the abundance of nitrogen cycling genes (reflecting the metabolic capacities) and the activity of ammonia oxidizers. Since effects on community composition and the composition of functional genes may not be detectable during short-time incubations, we chose to examine changes in functional gene expression (as a proxy for activity), namely, quantifying the expression of amoA genes (coding for ammonia monooxygenase) - genes involved in nitrification - which is a critical N cycling process facilitating N loss through coupled nitrification-denitrification in the Baltic Sea coastal zone (Hietanen et al., 2012). We anticipated temporal functional succession and treatment-specific responses to our manipulations, and further that effects would differ between the two localities with

a more modest response in the relatively DOM and nitrogen-rich northern locality.

Results and discussion Data presented here represent a subset of two larger experiments conducted with bacterial communities from the southern (Øresund, Exp I, April 2015) and the northern (Storfjärden, Gulf of Finland, Exp II, July 2015) Baltic Sea. Details on setup, sampling, and analyses are provided in Markussen et al. (submitted). Briefly, the microcosms consisted of three treatments in triplicates (control in 6 replicates) including 0.2 µm filtered seawater (control) amended with 20% vol/vol of 0.2 µm filtered humic river water (DOMhum) or agricultural river water (DOMagri). The salinity of the river water was adjusted with muffled NaCl to in situ levels of 13.4 (Øresund) and 6 (Storfjärden). A plankton inoculum (<10 µm) representing 20 % of the final volume was added and the microcosms were incubated in the dark at in situ temperature ±3°C for 5 (Exp I) or 4 days (Exp II). Daily samplings at 09.00 and 21.00 covered a variety of environmental parameters and DNA/RNA sampling.

Table 1. Concentrations of dissolved organic carbon (DOC), dissolved organic nitrogen (DON) ammonium (NH4), phosphate (PO4) and nitrate (NO3) in the treatments at the beginning of the experiments for the controls and the agriculture river water treatments (DOMagri) and the humic river water treatments (DOMhum). Standard deviations in brackets. Experimental procedures are described in Markussen et al. (submitted).

Exp I Exp II Treatment Control DOMhum DOMagri Control DOMhum DOMagri DOC (µM) 381.50

(13.66) 552.67 (51.43) 481.67 (78.23) 555.50 (80.77) 711.33 (66.98) 756.00 (59.43)

DON (µM) 47.37 (0.37) 51.97 (1.34) 103.57 (15.97) 48.05 (2.16) 51.77 (1.27) 57.10 (0.90) NH4 (µM) 0.57 (0.02) 0.94 (0.12) 1.03 (0.04) 0.35 (0.03) 0.54 (0.03) 1.12 (0.01) PO4 (µM) 0.06 (0.03) 0.20 (0.02) 0.19 (0.08) 0.09 (0.04) 0.08 (0.02) 0.09 (0.10) NO3 (µM) 1.74 (0.19) 3.63 (0.66) 61.87 (9.35) 0.27 (0.08) 2.60 (0.66) 0.56 (0.08)

83

The effect of agricultural and humic river water DOM on microbial communities In both experiments, there was a general increase in concentrations of DOC, DON, ammonium (NH4+) and nitrate (NO3-) in the DOMhum and DOMagri treatments relative to the controls. In particular high DON (104 µM) and NO3 (62 µM) concentrations were found in the DOMagri treatment in Exp I, whereas both river treatments had high DOC concentrations (711 and 756 µM, respectively; Table 1) in Exp II.

Bacterial production (BP) (Fig. 1A, C) and abundance (BA) (Fig. 1B, D) increased significantly over time in both experiments, but with large differences between the experiments. BA (both experiments) and BP (Exp II) were significantly increased in both the DOMagri and DOMhum treatments compared with the controls, while BP was

not significantly stimulated by either of the treatments in Exp I. The observed growth responses were anticipated to be accompanied by community dynamics mirrored in composition (16S rRNA genes) and function (composition and expression of N cycling genes), based on earlier studies reporting that DOM can shape bacterioplankton community composition (McCarren et al., 2010; Landa et al., 2015; Traving et al., 2017). However, a PCA revealed that community composition at the end of the experiments clustered into site rather than treatment (Fig. 2A). Moreover, the community composition differed significantly between Exp I and II (ANOSIM, r2=0.44, p<0.001) but not between treatments (ANOSIM, Exp I+II: r2=0.04, p=0.88, Exp I: r2=0.10, p=0.69, Exp II: r2=0.49, p=0.09). Similarly, Shannon diversity (r2 = 0.67, p < 0.001) and taxonomic richness was

Figure 1. Bacterial abundance (measured by flow cytometry) and bacterial production (measured by thymidine incorporation) during Exp I (A, B) and Exp II (C, D) for the controls and the agriculture (DOMagri) and the humic (DOMhum) river water treatments. Data are adapted from Markussen et al. (submitted).

84

Figure 2. Principal component analysis (PCA) of community composition (based on 16S rRNA sequencing; A) and nitrogen related genes (EC/eggNOG/PFAM) (metagenomics sequencing; B) at the end of Exp I and Exp II for the controls and the agriculture (DOMagri) and humic (DOMhum) river water treatments. For separate PCA of each experiment; see Figure S4. 16S rRNA genes were amplified using Bakt_341F and Bakt_805R (Herlemann et al., 2011) and sequenced on the Illumina MiSeq platform. Sequences were trimmed, quality filtered, and clustered (97% cut-off) into operational taxonomic units (OTUs) using Qiime (Caporaso et al., 2010) and the SILVA database (Pruesse et al., 2007). Sequences were deposited in NCBI SRA (542 PRJNA435478). Metagenomic libraries were sequenced on Illumina HiSeq2000 (125 bp) and quality trimmed using Cutadapt (Martin, 2011). Duplicate reads were removed using fastuniq (Xu et al., 2012). High quality reads were mapped using Bowtie2 (Langmead and Salzberg, 2012) against the BARM database (Alneberg et al., in press). Raw counts of BARM genes were calculated from Bedtools histograms (Quinlan, 2014), and translated to counts of gene functions; Pfam (Finn et al., 2016) and Clusters of Orthologous Groups (COG) (Galperin et al., 2015) by using the gene annotations of BARM (Alneberg et al., in press). Both metagenomic and amplicons sequences have been deposited in NCBI SRA (Bioproject number PRJNA435478).

significantly higher in Exp II compared to Exp I (r2 = 0.24, p < 0.001); however, no significant differences in alpha-diversity were observed between treatments from the same experiment. Hence, the changes in bacterial growth were only accompanied by limited shifts in community composition – and this was observed in both examined environments with marked differences in local community composition.

At the phylum level, e.g. γ-proteobacteria were significantly over-represented in the DOMagri treatment in Exp I and in both

DOMagri and DOMhum in Exp II (Fig. 3A). In Exp II, e.g. β-proteobacteria were stimulated in the DOMhum treatment relative to the controls. Such stimulation of β-proteobacteria by DOM has previously been observed for Baltic bacterioplankton (Kisand and Wikner, 2003; Traving et al., 2017). At the family level, there were several responses within Proteobacteria (Fig. 3B); e.g. Alteromonadaceae was more abundant in the DOMagri treatment relative to control in Exp I (DOMagri: 58%; Control: 25%). Within Bacteroidetes responses were limited (Fig.

85

Figure 3. Relative abundance of phyla/sub-phyla (A), families within Proteobacteria (B) and Bacteriodetes (C) at the end of Exp I and II for the controls and the agriculture river water (DOMagri) and the humic river water (DOMhum) treatments. Significant differential abundances for each group between treatments were tested using EdgeR (Robinson et al., 2010) and indicated for Exp I (*) and Exp II (°).

3C). Hence, some differences were observed in composition between treatments (Fig. 3), but overall changes were considerably less than the difference between environments (Fig. 2). There are examples of resistant microbial composition withstanding disturbance (e.g. Bowen et al., 2011); however, it may also be that DOC manipulations, as in the current study, only select for some taxa whereas the relative abundance of most taxa remain unchanged. Hence, it appears that overall community structure is a relatively poor predictor of the

bacterial growth response, as suggested by others (Dinasquet et al., 2013).

Relative abundance of nitrogen cycling genes It has been suggested that the key level at which to address the assembly and structure of bacterial communities is not taxonomy but rather the more functional level of genes (Burke et al., 2011; Krause et al., 2014). Moreover, since N availability affects N cycling genes (e.g. Zhang et al., 2013), we hypothesized that the high N concentrations in the added river water would elicit extensive and differential responses in the relative abundance of N cycling genes at the two sites. Nevertheless, as in the compositional analysis, a PCA of the relative abundance of N cycling genes (EC/EggNOG/PFAM; see Table S2) revealed a clustering according to site (ANOSIM, r2=0.1944, P=0.003) rather than treatment (ANOSIM, Exp I+II: r2=0.06, p=0.62, Exp I: r2=0.15, p=0.50, Exp II: r2=0.51, p=0.037) (Fig. 2B). Generalized linear models (GLM) showed that the relative abundances of N cycling genes did not correlate with any environmental parameters in Exp I. In Exp. II, on the other hand, there were significant correlations with NH4 (LR=1438.8, p=0.026), DOC (LR=1592.3, p=0.011), DON (LR=1508.8, p=0.012) and treatment (LR=2665.3, p=0.018). This suggests that while initial natural communities had a significant impact on the functional response to the river water amendments, addition of river water with a high DOC to DON ratio in Exp. II (Table 1)

86

also affected the abundance of N cycling genes.

The relative abundance of several specific

genes did differ significantly between controls and treatments, but, in line with the above GLM results, more were over- or under- represented in Exp II than Exp I (Fig. S3). This suggests that a universal response (across sites) in N cycling genes due to treatment alone was not the case, but rather that the community of the northern Baltic Sea (Storfjärden) was more responsive than that of the southern Baltic Sea (Øresund). Reasons for this may include multiple site characteristics or seasonality; however, it is noteworthy that DOC levels naturally, and in our experiment, are highest in the northern Baltic (Sandberg et al., 2004; Rowe et al., 2018). Bacterioplankton in this environment may, therefore, be particularly adapted and responsive to pulses of riverine DOM. In addition, the higher diversity and taxonomic richness of Storfjärden could possibly have benefited the responsiveness of this community.

In Exp I, a N2 fixation related gene (nifB, COG0535) and a nitrous oxide reduction gene (nosZ, PF13473) were over-represented in the DOMhum treatment relative to the control. Further, both ammonia transporters (COG0004, PF00909) and nitrite/nitrate reductases (PF07732, PF00394, PF04879) were under-represented (Fig. S3A). The DOMagri treatment did not have any significant effect on the relative abundance of N related genes (Fig. S3B). In Exp II, several EC/COG/PFAMs (DOMhum: 20; DOMagri: 13) differed significantly in relative abundance between treatments and controls (Fig. S3C,D). Among these, some of the N2 fixation related genes were significantly over-

represented in the DOMhum treatment (COG1348, PF00142, PF00148, EC 1.18.6.1) while one was under-represented (PF04055) (Fig. S3C). Most ammonia and nitrite/nitrate transporters were under-represented (PF00909, COG0004, PF07690) along with two urease genes (PF07969, PF01979). Both nitrate reductase (EC 1.7.99.4) and nitrous oxide reductases (COG4263, EC 1.7.2.4) were over-represented. In the DOMagri treatment, a single N2 fixation gene was under-represented (COG0535) while both nitrite/nitrate transporters (COG2223) and nitrous oxide reductases (COG4263, EC 1.7.2.4, EC 1.7.1.14) were over-represented (Fig. S3D).

While there were only few responses in N related genes to the DOMagri treatments, there were several overlaps in the response to the DOMhum treatment between the two experiments. The under-representation of ammonia channel protein AmtB and over-representation of N2 fixation genes in the DOMhum treatments, both point to possible N limitation (Carini et al., 2018). There were, however, no indications of N limitation when looking at the N:P ratios. The N:P ratios (calculated as (NH4+NO3)/PO4) were highest in the DOMagri treatments in Exp. I and in the DOMhum treatment in Exp. II. Further, the over-representations of nitrate reductases (Exp I) and nitrous oxide reductases (both experiments) indicate a promotion of some steps of the denitrification pathway by the addition of DOMhum. Denitrification in the Baltic Sea is known from anoxic zones of the water column (Dalsgaard et al., 2013) and from sediments (Silvennoinen et al., 2007).

Abundance and activity of ammonia oxidizing archaea (AOA) and bacteria (AOB)

87

To quantitatively assess the impact of the treatments on functional gene abundance and expression, digital droplet PCR (ddPCR) was used to enumerate amoA genes and transcripts of AOA and AOB in each experiment initially, after 44-45 h, and at the end of each experiment (Fig. 4). The ddPCR method was chosen because of its insensitivity to the presence of PCR inhibitors and documented applicability in marine samples (Lee et al., 2017) and for quantification of amoA genes (Dong et al., 2014). Although amoA gene and transcript

abundances of both AOA and AOB were dynamic over time, there was no significant treatment effect on amoA abundances. There were, however, significant differences between the amoA transcript abundances of DOMhum and control treatments in Exp I for both AOA (one-way ANOVA; F=11.7, p=0.011 and one-way ANOVA on ranks; Q=2.6, p=0.014) and AOB (one-way ANOVA on ranks; Q=2.32, p=0.024 and Q=2.55, P=0.032, respectively) at the beginning of the experiment.

Figure 4. Archaeal (AOA) and bacterial (AOB) amoA gene and transcript abundances at the start of the experiments (A,D), after 44-45 hours (B,E) and at the end of the experiments (C,F) for Exp I (left) and Exp II (right) for the controls and the agriculture (DOMagri) and humic (DOMhum) river water treatments. Note the different scales in (B). AmoA genes and transcript abundances were quantified using a BioRad digital droplet PCR (ddPCR) system and chemicals (EvaGreen) and primers amoA-1F/amoA-2R (Rotthauwe et al., 1997) and arch-amoAF/arch-amoAR (Francis et al., 2005). Reverse transcription (RT) was done using the gene specific primers, and the concentration of RNA and cDNA (measured by RiboGreen and Picogreen, respectively; Invitrogen) was used to calculate the efficiency of the RT reaction (RT factor), which was used to correct final values (Happel et al., in press.)

88

In both experiments, we found abundances of AOA (3.9 x 103 – 7.3 x 104 copies L-1) and AOB (4.2 x 103 – 5.4 x 104 copies L-1) in similar ranges, whereas amoA gene expression was dominated by AOB (up to 1.5 x 109 copies L-1 for AOB and 1.5 x 108 copies L-1 for AOA). The dominance of amoA transcripts from AOB matches findings in a recent Baltic Sea study (Happel et al., in press). Surprisingly, there was no significant stimulation of amoA gene abundances from AOA or AOB by the addition of riverine DOM in neither experiment and amoA gene expression was in fact reduced in the DOMhum treatments relative to controls. This could be interpreted as a sign of ammonia limitation (Carini et al., 2018), however, since amoA gene expression from AOB was negatively correlated with DOC in Exp I (Pearson correlation, r=-0.34, p=0.04), we speculate that ammonia oxidizers were hampered by the introduction of riverine DOC. It is conceivable that the sudden availability of labile riverine carbon (Sandberg et al., 2004; Rowe et al., 2018) is at odds with the chemolithoautotrophic life strategy of ammonia oxidizers (Strauss and Lamberti, 2000; Strauss et al., 2002).

Concluding remarks Despite that the addition of river water caused several folds increase in bacterial growth in both the Øresund and Storfjärden experiments, only specific sub-populations and N cycling processes were affected by the treatments, whereas overall community composition and the collective pool of examined N cycling genes remained relatively unaffected. This may support the notion that many bacterioplankton species are generalists and less responsive to transient changes in the DOC pool (Mou et

al., 2008), and that the linkage between identity and specialized DOC utilization is valid only within some phyla or among specific sub-populations (Dinasquet et al., 2013). Interestingly, treatment effects on nitrification were only observed at the expression level, and not the gene level. This observation underlines that functional responses in key N cycling processes in bacterioplankton may not always be accompanied by selection affecting community composition. Overall, the higher responsiveness of the community in Storfjärden to riverine DOM sources, coupled with projected increases in precipitation and outlet of allochthonous DOM (Andersson et al., 2015), may indicate that the coastal zones in the Northern Baltic Sea will undergo more dramatic future changes in N cycling regimes than the communities in the Øresund.

Acknowledgements This work resulted from the BONUS Blueprint project supported by BONUS (Art 185), funded jointly by the EU and the Danish Council for Independent Research, Estonian Research Council, Swedish Research Council FORMAS, and Academy of Finland. We kindly thank Tvärminne Zoological Station, University of Helsinki, and its personnel for providing infrastructural support. The authors declare no conflict of interest. References Alneberg, J., Sundh, J., Bennke, C., Beier, S.,

Lundin, D., Hugerth et al. (2018) BARM and BalticMicrobeDB, a reference metagenome and interface to meta-omic data for the Baltic Sea. Scientific Data, in press.

Andersson, A., Meier, H.E.M., Ripszam, M., Rowe, O., Wikner, J., Haglund, P. et al. (2015)

89

Projected future climate change and Baltic Sea ecosystem management. Ambio 44: S345-S356.

Azam, F., Fenchel, T., Field, J.G., Gray, J.S., Meyer-Reil, L.A., and Thingstad, F. (1983) The ecological role of water-column microbes in the sea. Mar Ecol Prog Ser 10: 257-263.

Bernes, C. (2005) Change Beneath the Surface. An In-Depth Look at Sweden's Marine Environment. Fälth & Hässler, Värnamo, Sweden Monitor 19: 38-47.

Bowen, J.L., Ward, B.B., Morrison, H.G., Hobbie, J.E., Valiela, I., Deegan, L.A., and Sogin, M.L. (2011) Microbial community composition in sediments resists perturbation by nutrient enrichment. The ISME Journal 5: 1540-1548.

Bristow, L.A., Mohr, W., Ahmerkamp, S., and Kuypers, M.M.M. (2017) Nutrients that limit growth in the ocean. Current Biology 27: R474-R478.

Bronk, D.A., See, J.H., Bradley, P., and Killberg, L. (2007) DON as a source of bioavailable nitrogen for phytoplankton. Biogeosciences 4: 283-296.

Burke, C., Steinberg, P., Rusch, D., Kjelleberg, S., and Thomas, T. (2011) Bacterial community assembly based on functional genes rather than species. Proc Natl Acad Sci U S A 108: 14288-14293.

Carini, P., Dupont, C.L., and Santoro, A.E. (2018) Patterns of thaumarchaeal gene expression in culture and diverse marine environments. Environmental Microbiology. Doi:10.1111/1462-2920.14107.

Dalsgaard, T., De Brabandere, L., and Hall, P.O.J. (2013) Denitrification in the water column of the central Baltic Sea. Geochimica et Cosmochimica Acta 106: 247-260.

Dinasquet, J., Kragh, T., Schroter, M.L., Sondergaard, M., and Riemann, L. (2013) Functional and compositional succession of bacterioplankton in response to a gradient in bioavailable dissolved organic carbon. Environ Microbiol 15: 2616-2628.

Falkowski, P.G., Fenchel, T., and Delong, E.F. (2008) The microbial engines that drive Earth's biogeochemical cycles. Science 320: 1034-1039.

Finn, R.D., Coggill, P., Eberhardt, R.Y., Eddy, S.R., Mistry, J., Mitchell, A.L. et al. (2016) The Pfam protein families database: towards a more sustainable future. Nucleic Acids Res 44: D279-285.

Francis, C.A., Roberts, K.J., Beman, J.M., Santoro, A.E., and Oakley, B.B. (2005) Ubiquity and

diversity of ammonia-oxidizing archaea in water columns and sediments of the ocean. Proceedings of the National Academy of Sciences of the United States of America 102: 14683-14688.

Galperin, M.Y., Makarova, K.S., Wolf, Y.I., and Koonin, E.V. (2015) Expanded microbial genome coverage and improved protein family annotation in the COG database. Nucleic Acids Res 43: D261-269.

Glibert, P.M., Wilkerson, F.P., Dugdale, R.C., Raven, J.A., Dupont, C.L., Leavitt, P.R. et al. (2016) Pluses and minuses of ammonium and nitrate uptake and assimilation by phytoplankton and implications for productivity and community composition, with emphasis on nitrogen-enriched conditions. Limnology and Oceanography 61: 165-197.

Happel, E., Bartl, I., Voss, M. and Riemann, L. (2018) Extensive nitrification and abundance of active bacterial ammonia oxidizers in two contrasting coastal systems of the Baltic Sea. Environmental Microbiology, in press. Doi:10.1111/1462-2920.14293

Herlemann, D.P., Labrenz, M., Jurgens, K., Bertilsson, S., Waniek, J.J., and Andersson, A.F. (2011) Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. Isme j 5: 1571-1579.

Hietanen, S., Jäntti, H., Buizert, C., Jürgens, K., Labrenz, M., Voss, M., and Kuparinen, J. (2012) Hypoxia and nitrogen processing in the Baltic Sea water column. Limnology and Oceanography 57: 325-337.

Kisand, V., and Wikner, J. (2003) Combining culture-dependent and -independent methodologies for estimation of richness of estuarine bacterioplankton consuming riverine dissolved organic matter. Applied and Environmental Microbiology 69: 3607-3616.

Knudsen-Leerbeck, H., Mantikci, M., Bentzon-Tilia, M., Traving, S.J., Riemann, L., Hansen, J.L.S., and Markager, S. (2017) Seasonal dynamics and bioavailability of dissolved organic matter in two contrasting temperate estuaries. Biogeochemistry 134: 217-236.

Krause, S., Le Roux, X., Niklaus, P.A., Van Bodegom, P.M., Lennon, J.T., Bertilsson, S. et al. (2014) Trait-based approaches for understanding microbial biodiversity and ecosystem functioning. Frontiers in Microbiology 5: 1-10.

Landa, M., Blain, S., Christaki, U., Monchy, S., and Obernosterer, I. (2015) Shifts in bacterial

90

community composition associated with increased carbon cycling in a mosaic of phytoplankton blooms. The Isme Journal 10: 39.

Langmead, B., and Salzberg, S.L. (2012) Fast gapped-read alignment with Bowtie 2. Nature Methods 9: 357.

Martin, M. (2011) Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnetjournal 17: 10.

McCarren, J., Becker, J.W., Repeta, D.J., Shi, Y., Young, C.R., Malmstrom, R.R. et al. (2010) Microbial community transcriptomes reveal microbes and metabolic pathways associated with dissolved organic matter turnover in the sea. Proceedings of the National Academy of Sciences 107: 16420-16427.

Mou, X., Sun, S., Edwards, R.A., Hodson, R.E., and Moran, M.A. (2008) Bacterial carbon processing by generalist species in the coastal ocean. Nature 451: 708-711.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B.M., Ludwig, W., Peplies, J., and Glöckner, F.O. (2007) SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Research 35: 7188-7196.

Quinlan, A.R. (2014) BEDTools: the Swiss-army tool for genome feature analysis. Current protocols in bioinformatics 47: 11.12.11-11.12.34.

Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010) edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26: 139-140.

Rönnberg, C., and Bonsdorff, E. (2004) Baltic Sea eutrophication: area-specific ecological consequences. Hydrobiologia 514: 227-241.

Rotthauwe, J.H., Witzel, K.P., and Liesack, W. (1997) The ammonia monooxygenase structural gene amoA as a functional marker: Molecular fine-scale analysis of natural ammonia-oxidizing populations. Applied and Environmental Microbiology 63: 4704-4712.

Rowe, O.F., Dinasquet, J., Paczkowska, J., Figueroa, D., Riemann, L., and Andersson, A. (2018) Major differences in dissolved organic matter characteristics and bacterial processing over an extensive brackish water gradient, the Baltic Sea. Marine Chemistry 202: 27-36.

Ryther, J.H., and Dunstan, W.M. (1971) Nitrogen, Phosphorus, and Eutrophication in the Coastal Marine Environment. Science 171: 1008-1013.

Sandberg, J., Andersson, A., Johansson, S., and Wikner, J. (2004) Pelagic food web structure and carbon budget in the northern Baltic Sea potential importance of terrigenous carbon. Marine Ecology Progress Series 268: 13-29.

Seitzinger, S.P., Harrison, J.A., Dumont, E., Beusen, A.H.W., and Bouwman, A.F. (2005) Sources and delivery of carbon, nitrogen, and phosphorus to the coastal zone: An overview of Global Nutrient Export from Watersheds (NEWS) models and their application. Global Biogeochemical Cycles 19: 1-11.

Silvennoinen, H., Hietanen, S., Liikanen, A., Stange, C.F., Russow, R., Kuparinen, J., and Martikainen, P.J. (2007) Denitrification in the river estuaries of the northern Baltic Sea. Ambio 36: 134-140.

Strauss, E.A., and Lamberti, G.A. (2000) Regulation of nitrification in aquatic sediments by organic carbon. Limnology and Oceanography 45: 1854-1859.

Strauss, E.A., Mitchell, N.L., and Lamberti, G.A. (2002) Factors regulating nitrification in aquatic sediments: effects of organic carbon, nitrogen availability, and pH. Canadian Journal of Fisheries and Aquatic Sciences 59: 554-563.

Traving, S.J., Rowe, O., Jakobsen, N.M., Sorensen, H., Dinasquet, J., Stedmon, C.A. et al. (2017) The Effect of Increased Loads of Dissolved Organic Matter on Estuarine Microbial Community Composition and Function. Front Microbiol 8: 351.

Xu, H., Luo, X., Qian, J., Pang, X., Song, J., Qian, G. et al. (2012) FastUniq: A Fast De Novo Duplicates Removal Tool for Paired Short Reads. PLOS ONE 7: e52249.

Zehr, J.P., and Ward, B.B. (2002) Nitrogen cycling in the ocean: new perspectives on processes and paradigms. Appl Environ Microbiol 68: 1015-1024.

Zhang, X., Liu, W., Schloter, M., Zhang, G., Chen, Q., Huang, J. et al. (2013) Response of the Abundance of Key Soil Microbial Nitrogen-Cycling Genes to Multi-Factorial Global Changes. PLOS ONE 8: e76500.

91

92

Supporting Information

93

94

Supporting information

Effects of allochthonous DOM input on microbial composition and nitrogen cycling genes at two contrasting estuarine sites Elisabeth M. Happel1, Trine Markussen1, Jonna E. Teikari2, Vimala Huchaiah3, Johannes Alneberg4, Anders F. Andersson4, Kaarina Sivonen2, Matthias Middelboe1, Veljo Kisand3, Lasse Riemann1*.

*Correspondence: Lasse Riemann, lriemann@bio.ku.dk

Figure S1. Map of the Baltic Sea showing the sampling sites of Exp I (Øresund) and Exp II (Storfjärden), and the locations of the humic water river (Lapväärti) and the agricultural water river (Lielupe).

95

Table S2. Table of N cycling processes, genes, E.C., COG, EgNOGG and PFAM numbers. The numbers used in the statistical analysis are marked in bold.

Process Protein gene EC EGNOGG COG PFAM

N2 is fixed into NH3 Nitrogenase iron subunit nifH 1.18.6.1 ENOG4105DSM COG1348 PF00142

N2 is fixed into NH3 Nitrogenase molybdenum-iron protein alpha chain (EC 1.18.6.1) (Dinitrogenase) (Nitrogenase component I) nifD ENOG4105D0B COG2710 PF00148

N2 is fixed into NH3 Nitrogenase iron-molybdenum cofactor biosynthesis protein NifE nifE - ENOG4105D0B COG2710 PF00148

N2 is fixed into NH3 FeMo cofactor biosynthesis protein NifB (EC 4.-.-.-) (Nitrogenase cofactor maturase NifB) (Radical SAM assemblase NifB) nifB - ENOG4105CGC COG0535 PF02579,PF04055

N2 is fixed into NH3 Nitrogenase molybdenum-iron protein beta chain (EC 1.18.6.1) (Dinitrogenase) (Nitrogenase component I) nifK ENOG4105CY9 COG2710 PF11844,PF00148

N2 is fixed into NH3 Nitrogenase iron-molybdenum cofactor biosynthesis protein NifN nifN - ENOG4105DK8 COG2710 PF00148

N2 is fixed into NH3 Nitrogenase-associated protein nifO - ENOG4105ND9,ENOG4111TNF PF03960

N2 is fixed into NH3Nitrogenase iron-molybdenum cofactor biosynthesis protein NifX (Nitrogenase iron-molybdenum cofactor biosynthesis protein NifX 1) (Nitrogenase iron-molybdenum cofactor biosynthesis protein NifX 2) nifX - ENOG4108UR2,ENOG4111JNF PF02579

N2 is fixed into NH3 anfG 1.19.6.1NH3 is oxidized into hydroxylamine Ammonia monooxygenase amoA 1.14.99.39 - PF12942Hydroxyolamine is oxidized into nitrite Hydroxylamine oxidase (cytochrome) hao 1.7.2.6

Nitrite is oxidized into nitrate Nitrite oxyreductase nxrAB 1.7.99.?

Nitrate is reduced into nitrite Nitrate reductase narG, NarZ 1.7.5.1

Nitrated is reduced into nitrite Ferredoxin nitrate reductase narB 1.7.7.2

Nitrate is reduced into ammonia Nitrate reductases1.7.1.1/1.7.1.2/1.7.1.3

Nitrite is reducted into nitric oxide Copper-containing nitrite reductase (EC 1.7.2.1) (Cu-NIR) nirK 1.7.2.1 ENOG4105CEI COG2132 PF00394,PF07732

nirS 1.7.2.1 ENOG4105F64 COG2010hydroxylamine is reduced to ammonia hydroxylamine reductase nirS 1.7.99.1 ENOG4105F64 COG2010 PF13442Nitric oxide is reduced into nitrous oxide Nitric oxide reductase norB 1.7.2.5Nitric oxide is reduced into nitrous oxide Nitric oxide reductase (fungal) CYP55 1.7.1.14

Nitrous oxide is reduced into N2 Nitrous oxide reductase nosZ 1.7.2.4 ENOG64105EQJ COG4263 PF13473Ammonia and hydroxilamine (from nitrite) forms hydrazine Hydrazine synthase 1.7.2.7

Hydrazine is formed into N2 Hydrazine dehydrogenase 1.7.2.8

Nitrate is reduced into nitrite Ferredoxin-nitrate reductase narB 1.7.7.2

Nitrate reductases1.7.1.1/1.7.1.2/1.7.1.3

Assimilatory nitrate reductase 1.7.99.?

Nitrite reductase 1.7.1.4

Ferredoxin-nitrite reductase nirA 1.7.7.1 ENOG4105ET1 COG0155 PF01077,PF03460

Nitrate is reduced into nitrite Nitrate reductase narG, narZ 1.7.5.1

Periplasmic nitrate reductasenapA, narB, nasA 1.7.99.4 ENOG4107QIW COG0243 PF04879,PF00384,PF01568,PF04324

Nitrite reductase (NADH) large subunit (EC 1.7.1.15) nirB 1.7.1.15 ENOG4107QZF COG1251 PF04324,PF01077,PF03460,PF07992

Nitrite reductase (cytochrome) nrfA 1.7.2.2 PF02335

ammonia trannsporter amt ENOG4105BZU,ENOG410XNMH PF00909

Ammonium transporter Amt amt ENOG4105BZU COG0004,COG0347 PF00909,PF00543

Ammonium transporter (Membrane protein NrgA) (Protein AmtB) nrgA ENOG4105BZU,ENOG410XNMH PF00909

Urea ABC transporter urtA ENOG4107RB5 COG0559 PF02653

Nitrate transport ATP-binding protein NrtC nrtC ENOG4105D7T COG0715,COG1116 PF00005

nrtA Synpcc7942_1239 nrtA ENOG4105DG7 COG0600 PF00528

nrtB Synpcc7942_1238 nrtB ENOG4105D7T COG0715,COG1116 PF00005

Nitrate transporter nasA PF07690

Nitrate/nitrite transporter NrtP (Nitrate/nitrite permease) nrtP ENOG4107R3J COG2223 PF07690

Urease subunit alpha (EC 3.5.1.5) (Urea amidohydrolase subunit alpha) ureA 3.5.1.5 ENOG4108YZ9 COG0831,COG0832 PF00699

Global nitrogen regulator ntcA ENOG4107RU9,ENOG410XVK4 PF00027,PF13545

Glutamine synthetase (GS) (EC 6.3.1.2) (Glutamate--ammonia ligase) (Glutamine synthetase I alpha) (GSI alpha) glnA 6.3.1.2 ENOG4105C5F COG0174 PF00120,PF03951

Nitrogen regulatory protein P-II 1 glnB COG0347 PF00543

Urease subunit beta (EC 3.5.1.5) (Urea amidohydrolase subunit beta) ureB 3.5.1.5 ENOG4105CQM COG0804 PF01979,PF00449

Urease accessory protein UreD ureD1 COG0829 PF01774

Urease accessory protein UreD ureD ENOG4108M9X COG0829 PF01774

Urease accessory protein UreE ureE ENOG4105HFU COG2371 PF05194,PF02814

Urease accessory protein UreF ureF ENOG4108YH3 COG0830 PF01730

Urease accessory protein UreG ureG ENOG4105E8T COG0378 PF02492

Urease accessory protein UreH ureH ENOG4108XNJ COG0829 PF01774

UreX ureX ENOG4108U78 COG0625 PF13409

Urease subunit (EC 3.5.1.5) ureX 3.5.1.5 PF07969

96

Figure S3. Volcano plots showing up- and down- represented EC/PFAM/COGs in Exp I (A, B) and Exp II (C, D) for the controls and the humic (DOMhum)(A, C) and the agriculture (DOMagri) (B, D) river water treatments. Differential abundance analysis was done using EdgeR (Robinson et al., 2010). Only differentially abundant EC/PFAM/COGs with p<0.01 and FDR<0.05 were considered significant (black dots).

97

Figure S4. Principal component analysis (PCA) of OTUs (based on 16S rRNA; Exp I: A and Exp II: B) and N cycling genes (metagenomes; Exp I: C and Exp II: D).

98

Paper III

99

100

Extensive nitrification and active ammonia oxidizers in two contrasting coastal systems of the Baltic Sea

Elisabeth Happel1, Ines Bartl2, Maren Voss2, Lasse Riemann1*

1Marine Biological Section, Department of Biology, University of Copenhagen, Helsingør, Denmark

2Department of Biological Oceanography, Leibniz Institute for Baltic Sea Research (IOW), Rostock, Germany

*Corresponding author: Lasse Riemann (lriemann@bio.ku.dk), Strandpromenaden 5, 3000 Helsingør (DK), phone +4535321959.

Nitrification is important in nitrogen (N) cycling of aquatic environments, but knowledge about its regulation and importance is sparse. Here we examined nitrification and ammonia oxidizers in the Baltic Sea. We investigated two sites with different catchment characteristics (agricultural and forest), the Bay of Gdánsk (south) and the Öre Estuary (north), and measured pelagic nitrification rates and abundance, composition, and expression of ammonia monooxygenase (amoA) genes. Highest nitrification rates were found in the nutrient rich Bay of Gdańsk. Interestingly, abundances of ammonia-oxidizing archaea (AOA) and bacteria (AOB) were orders of magnitude lower than reported from other sites. Although AOA were most abundant at both sites, the highest expression levels were from AOB. Interestingly, few AOA and AOB taxa dominated amoA gene expression, with a Nitrosomarinus related phylotype showing widespread expression. AOA and AOB communities differed between sites and depths, respectively, with the composition in rivers being distinct. A storm event, causing an even depth distribution of nitrification and particles in the Bay of Gdańsk, indicated that the presence of particles stimulate nitrification. The study highlights coastal regions as dynamic sites of extensive pelagic nitrification, which may affect local food web dynamics and loss of N mediated by denitrification.

Introduction

Nitrification, the stepwise oxidation of ammonia to nitrite and ultimately to nitrate, is important in the marine nitrogen (N) cycle. Although nitrification does not directly

influence the system inventory of N, it alters the presence of N from its most reduced form, ammonia (NH3), to its most oxidized form, nitrate (NO3-) (Ward, 2005). This may affect not only the productivity of the system but also the composition of the phytoplankton

101

and the microbial community since different groups are known to have preferences for nitrate or ammonium, respectively (Ward, 2013; Glibert et al., 2016; Goldberg et al., 2017). Further, nitrate is important as substrate for denitrification, and nitrification may therefore indirectly facilitate N loss (Ward, 2013). Estuaries and coastal zones are generally highly productive (Gattuso et al., 1998) and often severely influenced by anthropogenic activities on land. For instance, the use of agricultural fertilizers in recent decades has increased loads of reduced N compounds to estuaries worldwide and stimulated the productivity of these vulnerable systems (Glibert et al., 2016). In particular in such nutrient-rich waters, the coupled nitrification-denitrification may be critical for N removal, counteract eutrophication (Hietanen et al., 2012), and constrain export of bioavailable N to the open sea. Hence, insight into nitrification and its regulation in estuaries is a prerequisite for a mechanistic understanding of local as well as regional nutrient and carbon fluxes and pools. Rates of nitrification in coastal areas vary in space and time. They can peak at the bottom of the euphotic zone (Newell et al., 2013), or in the bottom waters (Bristow et al., 2015) where competition with phytoplankton for ammonia is low and light-inhibition of ammonia oxidizers is minimal. High nitrification rates are often found in turbid, nutrient-rich coastal surface waters, like river plumes (Bianchi et al., 1999), and extensive seasonality in nitrification occurs in temperate coastal zones (Andersson et al., 2006). Nevertheless, although intensively studied, there is no clear understanding of which environmental factors govern coastal nitrification (Damashek et al., 2016)

Ammonia oxidation is considered the rate-limiting step of nitrification and is carried out by a few groups of ammonia oxidizing bacteria (AOB) and ammonia oxidizing archaea (AOA) (Ward, 2013). The distribution and relative importance of AOA and AOB is, however, poorly understood. Although AOA occur in both waters and sediments, coastal zones and open oceans, they often show a high degree of endemism, with only few groups being cosmopolitan (Francis et al., 2005). Interestingly, the first isolated AOA of the genus thaumarchaeota, Nitrosopumilus maritimus, turned out to have different biochemical pathways than AOB providing an advantage at low ammonium and oxygen conditions (Martens-Habbena et al., 2009), and pointing to discrete ecophysiologies of AOB and AOA. In the open ocean, AOA often dominate in abundance relative to AOB (Mincer et al., 2007; Beman et al., 2010). Likewise, AOA tend to dominate in coastal waters (Hollibaugh et al., 2011; Horak et al., 2013; Smith et al., 2014; Zhang et al., 2014) whereas AOB tend to dominate the sediment communities (Wankel et al., 2011; Zheng et al., 2014), with few exceptions (Abell et al., 2009).

The Baltic Sea is the second largest estuarine system in the world. It encompasses separate sub-basins with unique geology (Rönnberg and Bonsdorff, 2004), has catchment areas dominated by agriculture in the south and forestry in the north. Further, nutrient limitation of plankton production is changing from N limitation in the central Baltic Sea proper in the south to phosphorous limitation in the north (Bernes, 2005). In the present study, we sampled the Bay of Gdańsk (south) and the Öre Estuary (north) to gain first insights into nitrification and composition and activity of ammonia oxidizers in the Baltic Sea. While sampling

102

Cruise Date Station Depth (m)

Salinity Temp (°C)

Ammonium (µmol L-1)

Nitrate (µmol L-

1)

PON (µmol L-

1)

POC (µmol

L-1)

Chl. a

(µg L-1)

Gdańsk Feb. 9, 2015

Vistula River 0.30 0.10 1.20 11.96 178.11 5.69 54.66 3.11

Feb. 3, 2015

VE02 (before storm)

1.00 5.27 3.00 1.20 23.15 1.50 13.28 0.82

7.50 7.32 3.85 0.44 8.31 1.11 10.12 0.37 21.25 7.58 4.40 0.45 4.00 0.86 8.01 0.26 Feb. 4,

2015 22.10 7.53 NA 0.31 4.33 3.26 32.69 0.43

Gdańsk Feb. 9, 2015

VE02 (after storm)

2.50 7.49 3.34 0.32 5.02 4.10 30.54 3.18

7.50 7.49 3.34 0.39 5.05 3.79 29.99 3.13 18.00 7.54 3.33 0.30 4.71 4.52 36.64 3.58

Öre (April)

April 23-24, 2015

B1 (river) 0.30 0.01 3.00 0.68 5.75 12.28 153.56 NA

N6 16.05 5.15 3.13 0.17 0.71 4.96 39.89 6.99 N11 22.80 5.31 2.77 0.16 1.81 4.89 34.77 6.36 NB8 33.30 4.32 2.66 0.07 2.56 3.82 33.28 3.05 N14 17.00 5.35 2.75 0.06 3.96 2.43 19.35 2.38

Öre (August)

Aug. 5-7, 2015

B1 (river) 0.30 0.02 17.48 0.33 0.48 5.53 67.24 NA

N6 16.65 4.52 8.51 1.11 1.24 1.81 17.75 0.61 N14 30.15 4.80 7.03 0.48 2.25 2.03 2.03 NA

Table 1. Physical, chemical, and biological variables of the sampled stations in the Bay of Gdańsk, the Vistula River, and the Öre Estuary and river. Particulate organic nitrogen (PON), particulate organic carbon (POC), and chlorophyll a (chl.a). For positions of stations, see Fig. 1.

the Bay of Gdańsk, a storm gave us the opportunity to investigate short-term effects of water column mixing on nitrification and the composition of ammonia oxidizers. Results Samples were collected during three cruises in the Baltic Sea in 2015 (Fig. 1, Table 1): a) February in the Bay of Gdańsk and Vistula River, before and immediately after a storm; b) April and August from five stations in a transect from the Öre River (station B1) to more open waters of the Öre Estuary (Bothnian Sea). Samples for measurements of environmental variables and nitrification were taken from the surface, bottom water,

and occasionally from intermediate depths. Based on measurable nitrification, samples were selected for sequencing and quantification of ammonia monooxygenase subunit A (amoA) genes. Environmental variables Environmental variables are summarized in Table 1. The salinity in the Bay of Gdańsk (5.27 - 7.58) was significantly higher than in Öre Estuary (4.32 - 5.32) (p=0.004), and close to zero in the Vistula and Öre rivers. The temperature in the Bay of Gdańsk in February (3.00 - 4.40°C) was comparable to the Öre Estuary in April (2.60 - 3.13°C). In the

103

Figure 1. Map showing the Baltic Sea (A) and the sampling stations in the Bay of Gdańsk (B) and Öre Estuary (C).

Öre area the temperature increased significantly (p=0.044) from April to August. Overall, nitrate and ammonium concentrations were highest in the Bay of Gdańsk, especially in winter, whereas particulate organic nitrogen (PON) concentrations were highest in the Öre Estuary. The Vistula River had the highest ammonium (11.96 µmol L-1), nitrate (187.11 µmol L-1), and PON (5.96 µmol L-1) concentrations. Before the storm, ammonium (1.20 µmol L-1) and nitrate (23.15 µmol L-1) concentrations in the Bay of Gdańsk (VE02, close to the mouth of the Vistula River) were highest at the surface (1 m), whereas PON (3.26 µmol L-1) and particulate organic carbon (POC, 32.69 µmol L-1) concentrations were highest at depth (22 m). PON and POC were even higher in the Vistula River. Chlorophyll a (chl. a) was 3.11 µg L-1 in the Vistula River and 0.26 - 0.82 µg L-1 in the Bay of Gdańsk. After the storm, ammonium, nitrate, PON, and POC were similar throughout the water

column, and chl. a had increased throughout the water column to 3.13 - 3.58 µg L-1. In the Öre River in April, ammonium (<1 µmol L-1) and nitrate (5.75 µmol L-1) concentrations were higher than in the Öre Estuary, but roughly an order of magnitude lower than in the Vistula River. However, the PON concentration (12.28 µmol L-1) was higher than in the Vistula River (5 µmol L-1). Ammonium and PON concentrations decreased slightly from the Öre River mouth (N6) towards the open water, whereas the opposite was observed for nitrate. In August, concentrations of ammonium (p=0.020) and nitrate (p<0.001) were significantly lower than in April. The highest POC concentration in the Öre region was in the river in April (153.56 µmol L-1) and in August (67.24 µmol L-1), whereas chl. a was highest at the two outer stations (NB8 and N14) in April.

104

Nitrification Nitrification was generally higher in the Bay of Gdańsk than in Öre Estuary. A storm in the Bay of Gdańsk with up to 10 Bft (Beaufort scale, or 25 km s-1) allowed us to examine effects of mixing induced by the storm. The storm strongly affected the depth distribution of nitrification (Fig. 2). Before the storm, nitrification was lowest at the surface (47 nmol L-1 d-1 at 1m) and higher with depth (219 nmol L-1 d-1 at 21.25 m), but after the storm surface and mid-water rates had increased. The redistribution of nitrification in the water column did, however, only have a minor effect on the areal nitrification as approximated by trapezoidal integration (2.35 mmol m-2 d-1 before the storm vs. 2.86 mmol m-2 d-1 after the storm).

Figure 2. Archaeal (A) and bacterial (B) amoA gene abundance (DNA) and transcript abundances (RNA) in the Bay of Gdańsk (station VE02). Samples were taken before (BS) and after (AS) a storm. Lines showing nitrification rates (NR). Nitrification was not measured in the Vistula River.

In the Öre region, the lowest nitrification was in the river (10 nmol L-1 d-1) in April (Fig. 3) whereas the highest rate was at the outermost station N14 (39 nmol L-1 d-1). In August, nitrification was highest at the N6 station closest to the river; unfortunately nitrification was not measured in the river. Nitrification did not correlate with any of the measured environmental variables in either Bay of Gdańsk or Öre Estuary. Abundances of amoA genes and transcripts Digital droplet PCR (ddPCR) was used to quantify abundances of amoA genes and transcripts from AOA and AOB in 8 samples from each site (16 samples in total). In the Bay of Gdańsk AOA abundances exceeded that of AOB by up to an order of magnitude at all depths (Fig. 2). The abundance of AOA was highest in the Vistula River (5.8 x 104 copies L-1) and before the storm at the surface (4.1 x 104 copies L-1), decreasing with depth (Fig. 2A). In contrast, AOA transcript abundances were lowest at the surface (~4.0 x 104 copies L-1) and highest at 7.5 m depth (~1.6 x 106 copies L-1). After the storm, AOA gene and transcript abundances had decreased, except at the surface where there was a slight increase for transcripts. Although the AOA genes were more abundant than AOB genes, AOB transcript abundance greatly exceeded that of AOA, with highest abundances at 7.5 m (9.2 x 107

105

Figure 3. Archaeal (A) and bacterial (B) amoA gene abundance (DNA) and transcript abundances (RNA). Stations represent a transect from the river (B1) to the open sea (N14). Lines showing nitrification rates (NR). Samples were taken in April and August.

copies L-1) and at 21 m (6.3 x 107 copies L-1; Fig. 2B). In Öre Estuary, AOA gene abundance exceeded that of AOB at all stations (Fig. 3). AOA gene abundances (Fig. 3A) were highest in the river (B1) in April (9.7 x 104 copies L-1) and August (2.9 x 104 copies L-1), and decreased towards the outer station (N14). In contrast, AOA transcripts where lowest in the river (1.7 x 104 copies L-1) and increased towards the outer station (~3.1 x 106 copies L-

1). AOA gene and transcript abundances were generally higher in April than in August, at all stations. AOB transcript abundance was generally higher than that of AOA, ranging from ~6.2 x 104 copies L-1 in August to 7.8 x 108 copies L-1 in April at the outer station (N14; Fig. 3B).

In the Bay of Gdańsk AOA gene abundances were negatively correlated with salinity (Pearson’s coefficient, r = -0.86, p = 0.0037) and positively correlated with ammonium (p = 0.018, r = 0.78) and nitrate (r = 0.80, p = 0.014) concentrations (Supporting Information Figure S3). AOB gene abundances were not correlated with any environmental variables. In Öre Estuary, AOA gene abundances were positively correlated with PON (r = 0.95, p <0.0001) and POC (r = 0.98, p = 0.0002). Further, AOA gene abundances were negatively correlated with depth (r = -0.72, p = 0.045) and salinity (r = -0.79, p = 0.024). AOB gene abundances were positively correlated with temperature (r = 0.87, p = 0.016).

106

Figure 4. Principal component analysis (PCA) of the ammonia oxidizing archaeal (A) and bacterial (B) communities in the Bay of Gdańsk (incl. Vistula River) and Öre Estuary (incl. river). River samples are encircled for illustrative purposes. Light symbols: Öre River and Estuary; dark symbols: Gdańsk River and Bay. Community composition of ammonia oxidizers

In order to examine composition of total and active AOA and AOB amoA genes were Illumina sequenced from PCR amplified DNA and RNA (cDNA) samples, selected based on presence of measurable nitrification. Archaeal amoA genes were amplifiable from 7 samples

from the Bay of Gdańsk (gene transcripts from 6 samples) and 7 samples from the Öre Estuary (gene transcripts from 3 samples). Bacterial amoA genes were amplified from 7 samples from the Bay of Gdańsk and from 6 samples from the Öre Estuary. Unfortunately, AOB amoA transcripts could only be amplified by end-point PCR from one sample (see Experimental procedures). This means that sequence information on active AOB is only available from this sample. Sequencing yielded 457,530 and 2,917,947 archaeal and bacterial amoA gene reads, respectively. After quality filtering and chimera check, reads were clustered at 97% sequence similarity, forming 298 AOA operational taxonomic units (OTUs) and 169 AOB OTUs. Principal component analysis (PCA) was performed to compare the overall composition of AOA and AOB between sites (Fig. 4). Further, the relative abundance of the most abundant AOA and AOB OTUs (each OTU representing >0.6% of the total number of reads) were compared to identify key OTUs and compositional differences between samples (Fig. 5 and 6).

AOA communities of the Vistula and Öre rivers were significantly (p=0.006) distinct from the estuarine communities, while the communities in the two estuaries overlapped (Fig. 4). The AOB communities of the rivers were, however, not significantly (p=0.136) different from the estuarine communities. Indeed, several of the most abundant archaeal ammonia oxidizers (aOTU) were present in both the Bay of Gdańsk and Öre Estuary, with a few OTUs appearing exclusively at one site (e.g. aOTUs16, 17 and 19; Fig. 5). The Vistula and Öre rivers were dominated by aOTU4, aOTU8 and aOTU11, respectively. aOTU4 was identical to sequences from a lake sediment, aOTU8 showed 99% similarity to a sequence

107

from lake water, and aOTU11 was 99% similar to sequences from soil. Although the composition of the total community (DNA) differed between sites, the composition of the active community (RNA) showed almost no variation across sites and samples (Fig. 5). The dominant aOTU1 was present (DNA) at both sites (including rivers), with relative abundances ranging from < 5% in the rivers to almost 50% in the Bay of Gdańsk (at 21 m). Interestingly, aOTU1 represented roughly 80% of the active fraction of the communities, making it by far the most active OTU. It was phylogenetically associated with Candidatus Nitrosomarinus catalina SPOT01 (Ahlgren et al., 2017) (94.7% similarity) whereas aOTU2, the most dominant OTU in the Bay of Gdańsk (after the storm) and at the outer stations in Öre Estuary (NB8 and N14), was identical to a

sequence obtained from Baltic Sea sediments. Several AOA OTUs clustered with sequences from environments around the globe (Supporting Information Figure S1). Interestingly, aOTU2 was not abundant in the active fraction of the community. In the Bay of Gdańsk (before the storm), there was a change in AOA and AOB composition with depth (Fig. 5A and 6A). Several archaeal OTUs (e.g. aOTU4, aOTU6, aOTU12, aOTU14) were shared between the river and the shallow samples (0 m and 7.5 m), and not present at depth. The shallow sample (7.5 m) was dominated by two bacterial OTUs (bOTU1 and bOTU2) and the number of OTUs making up the communities increased with depth. The storm affected the composition of AOB. Before the storm,

Figure 5. Composition of the most abundant OTUs among the total (DNA) and the active (RNA) archaeal ammonia oxidizers in the Bay of Gdańsk (A) and the Öre (B) regions. Sampling depths in the Öre region are provided in Table 1. The plot includes OTUs each accounting for >0.6% of the total number of reads.

108

several OTUs (e.g. bOTU5 to 9) where found mostly at depth (21 - 22 m; Fig. 6A). After the storm, these OTUs together made up ~50% of the community throughout the water column. Also the storm appeared to mix archaeal aOTUs 1 and 2 from depth (22 m) and up in the water column (7.5 and 18 m after the storm; Fig. 5A). Interestingly, the composition of the dominant active AOA was unaffected by the storm, and remained the same throughout the water column. The most abundant bacterial ammonia oxidizers (bOTU) in the Vistula and Öre rivers were dominated by the same two OTUs: bOTU3 and bOTU4 in the Öre and Vistula River, respectively (Fig. 6). bOTU 3 and bOTU4 clustered with Nitrosomonas (80% similar to N. cryotolerans) and Nitrosospira (no close match), respectively (Supporting Information Figure S2). The communities in the Bay of Gdańsk and Öre Estuary were also dominated by the same two OTUs (bOTU1, bOTU2). bOTU1 was identical to a sequence from Baltic Sea sediment and together with

bOTU2 clustered with Nitrosomonas (79% similarity to N. cryotolerans). Unfortunately, bacterial amoA transcripts could only be amplified from a single sample (7.5 m before the storm, see Experimental Procedures), and was dominated by the same OTUs as the total AOB community. To identify potential linkages between the composition of ammonia oxidizers and environmental variables a redundancy analysis (RDA) with permutations was applied. It showed that AOA composition was significantly correlated with PON (p=0.002, explaining 11.3 % of the variation), nitrate (p=0.017, explaining 6.8 % of the variation), and temperature (p=0.023, explaining 7.7 % of the variation). The AOB community was correlated with PON (p=0.001, explaining 27% of the variation), temperature (p=0.023, explaining 8.4% of the variation), ammonium (p=0.002, explaining 15.1% of the variation), and nitrification rates (p=0.042, explaining 5.3 % of the variation) (Supporting Information Figure S4).

Figure 6. Composition of the most abundant OTUs among the total (DNA) and active (RNA) bacterial ammonia oxidizers in the Bay of Gdańsk (A) and the Öre (B) regions. Note that bacterial amoA transcripts could only be successfully sequenced from one sample. Sampling depths in the Öre region are provided in Table 1. The plot includes OTUs each accounting for >0.6% of the total number of reads. Asterisk indicates the single RNA sample from which AOB could be amplified.

109

Discussion Nitrification The nitrification rates measured in the Bay of Gdańsk (~50-200 nmol L-1 d-1) and Öre Estuary (< 50 nmol L-1 d-1) are much higher than those reported from open ocean waters (~0 - 10 nmol L-1 d-1) (Clark et al., 2008; Shiozaki et al., 2016) but are comparable to rates in marine coastal zones, like San Francisco Bay (Damashek et al., 2016), and in deeper waters of the central Baltic Sea; e.g. 1-280 nmol L-1 d-1, with highest rates below the halocline (Enoksson, 1986), and 85 nmol L-1 d-1 in the oxic-anoxic interface at ~115 m depth (Hietanen et al., 2012). As suggested for marine regions (Ward, 2013; Glibert et al., 2016), ammonia oxidizers may compete with phytoplankton for ammonia and extensive nitrification may affect the competitive edge of phytoplankton taxa with preference for nitrate or ammonium, respectively, and at the same time support local denitrification (Silvennoinen et al., 2007). Hence, both food web dynamics and N removal in the Baltic Sea is likely influenced by extensive coastal nitrification. The level of nitrification was higher at station VE02 in the Bay of Gdańsk than in Öre Estuary, possibly due to high inorganic nitrogen loads from the Vistula River (154 µmol L-1 in February 2015 of which 5% was NH4+; data from the Polish monitoring Program). However, a comparison between the Bay of Gdańsk (covering numerous stations) and the Öre Estuary did not identify significant differences in coastal nitrification (Bartl et al., unpublished). Nitrification was not correlated with ambient ammonium concentration. Such linkage is reported for cultivated organisms (e.g. Ward, 1990) and some coastal zones (e.g. Bianchi et al., 1999). Environmental data are, however, often

ambiguous (Ward and Kilpatrick, 1990; Damashek et al., 2016; Shiozaki et al., 2016) probably because ammonium is rapidly taken up by phytoplankton (Twomey et al., 2005). The storm in the Bay of Gdańsk could provide clues to the environmental regulation of nitrification in this area. The complete mixing of the water column, evident from the even distribution of salinity, temperature, PON and POC with depth, caused enhanced nitrification rates in the upper waters while leaving depth integrated nitrification relatively unchanged. At the same time, major compositional changes in the ammonia oxidizing community in the upper water was induced while the composition of the active AOA did not change. Hence, the elevated nitrification in the upper waters may not be attributable to the introduction of putative ammonia oxidizing taxa from depth. Likewise, an introduction of ammonia from depth is unlikely due to its low concentration in the bottom water (<0.5 µmol L-1) before the storm. On the other hand, a high concentration of particles was transported from bottom waters and sediments to the upper water column and showed a vertical distribution similar to nitrification after the storm (POC and PON data, Table 1). Before the storm, nitrification increased with depth, a pattern consistent with observations from Monterey Bay (Ward, 2005) and San Francisco Bay (Damashek et al., 2016), reflecting the vertical distribution of particles. Previous studies have indicated that nitrification might be associated to particles (Karl et al., 1984), being local hotspots of microbial activity (Simon et al., 2002) and ammonium production (Ploug and Bergkvist, 2015). Based on these findings we hypothesized that particles getting distributed throughout the water column in the event of

110

extensive vertical mixing; e.g. induced by the storm in the Bay of Gdańsk, caused the increased nitrification rates observed after the storm. This could be by particle-associated nitrification and/or by remineralization of particles making ammonium available to nitrifiers in the surrounding water. We did, however, not find a statistical linkage between nitrification and PON or POC concentrations – probably due to our limited dataset - but this was recently established in a larger dataset from Bay of Gdańsk (Bartl et al., accepted). Likewise, correlation between nitrification and suspended particulate matter has been established in the San Francisco Bay (Damashek et al., 2016). While particle-associated nitrification remains speculative based on our data, it is noteworthy that aOTU1, which accounted for roughly 80% of the AOA amoA transcripts in our study, is closely related (89.9% nucleotide similarity) to a Nitrosopumilus strain (NF5) from the Northern Adriatic Sea exhibiting functional traits suggesting it can seek favorable microenvironments such as nutrient-rich particles (Bayer et al., 2015). Hence, indications of particle-associated nitrification exist but focused experimental work is needed to firmly test this idea. Abundance and distribution of ammonia oxidizers Concentrations of AOA and AOB (~104 - 105 gene copies L-1) in our samples were surprisingly low compared to data from other regions. For instance, the AOA abundances in the Bay of Gdańsk and Öre Estuary were markedly lower than the up to ca. 108 copies L-

1 found in the Gulf of California (Beman et al., 2008), the Atlantic Ocean (Wuchter et al., 2006), or the Southeastern US Coast (Hollibaugh et al., 2011), but similar to the range reported from the Yangtze River

estuary (Zhang et al., 2014). We applied the ddPCR technique for gene quantification because it is relatively insensitive to the existence of PCR inhibitors, has been shown to produce correct quantification of cells in marine samples (Lee et al., 2017), a verified protocol for amoA gene quantification was available (Dong et al., 2014), and our specific hardware has earlier been used successfully for gene quantification (e.g. Nygaard et al., 2016). Hence, we find a systematic methodological underestimation of amoA genes in our study unlikely, and therefore speculate whether these low abundances could be due to competitive exclusion imposed by the special Baltic Sea environment; e.g. by the sharp salinity gradients (Herlemann et al., 2016) or the massive outlet of terrigenous carbon in the Öre Estuary (Sandberg et al., 2004). We found higher amoA gene abundances of AOA (up to ~ 105 copies L-1) than AOB (up to ~3 x 104 copies L-1) across sites, stations and season (except B1 in August). Consistent with this AOA abundance commonly exceeds that of AOB in estuarine (Caffrey et al., 2007; Zhang et al., 2014) and open ocean (Wuchter et al., 2006; Mincer et al., 2007) waters, but the ratio is often variable (Wuchter et al., 2006; Beman et al., 2010). Indeed, a prominent role of AOA among ammonia oxidizers was anticipated given previous reports of their numerical (Labrenz et al., 2010) and activity predominance (Berg et al., 2015) in hypoxic Baltic waters. Despite that AOA were more abundant than AOB, we found higher abundances of AOB amoA transcripts (up to 7.8 x 108 transcripts L-1), particularly in the Öre Estuary. Dominance in abundance of AOB was previously documented from estuarine sediment (Wankel et al., 2011), sulfurous lakes

111

(Llorens-Mares et al., 2015) and microbial mats (Fan et al., 2015). However, to our knowledge, our study is the first to document high amoA gene transcripts from AOB in an estuary. Since the abundance of amoA gene transcripts from AOB (or AOA) did not correlate with any measured environmental variables, our study does not reveal the factors regulating ammonia oxidation per se; however, the statistical analyses do provide some relevant indications on regulation of the ammonia oxidizing community. AOA abundances were negatively correlated with salinity at both sites and positively correlated with ammonium and nitrate concentrations in the Bay of Gdańsk and POC and PON in Öre Estuary. Indeed, the highest AOA abundances were in the rivers and surface waters, and predominant OTUs (e.g. aOTU8) in the rivers were closely related to sequences from freshwaters (Supporting Information Figure S1). Hence, we speculate that Baltic AOA prefer freshwater, but a more comprehensive dataset is needed to verify this idea. Composition of ammonia oxidizers Next generation sequencing revealed the existence of diverse AOA and AOB assemblages in Baltic Sea waters. The diverse assemblage of AOA is in marked contrast to the reported predominance of a single AOA group, related to Nitrosopumilus maritimus, at the Baltic Sea halocline (Labrenz et al., 2010). The close resemblance of AOA OTUs to sequences found in discrete localities around the globe (Supporting Information Figure S1), points to a wide distribution of many of the AOA phylotypes of the Baltic Sea coastal zone. The high relative abundance and activity of aOTU1, related to Candidatus Nitrosomarinus catalina recently isolated from off the coast of California (Ahlgren et al., 2017), is one such

example. The AOB consisted of the genera Nitrosomonas and Nitrosospira. Interestingly, Nitrosomonas-like OTUs were present in both the Bay of Gdańsk (incl. Vistula River) and Öre Estuary (incl. Öre River), whereas Nitrosospira-like OTUs were confined to the rivers, indicating a preference for freshwater. Both genera have, however, previously been found in Baltic Sea waters and sediments (Kim et al., 2008; Vetterli et al., 2016), and in sediments (Chang et al., 2017) and waters (Bano and Hollibaugh, 2000; O'Mullan and Ward, 2005; Kim et al., 2008) from multiple localities. It is not understood what governs the distribution of the two genera, possibly as a consequence of difficulties associated with correlating single environmental factors to complex combinations of variables shaping AOB diversity in dynamic environments (Francis et al., 2003). Despite of widespread amoA gene expression of AOB in our samples, as shown by ddPCR, sequencing was only successful for a surface sample from the Bay of Gdańsk (see experimental procedures). This sample suggested that the main present OTUs were also the most active. In contrast, AOA amoA transcript sequences from many samples showed that the composition of AOA transcripts was different to that of AOA genes, and almost identical across sites and samples, despite pronounced differences in AOA composition. The most dominant AOA OTU in DNA (aOTU1) represented roughly 80% of the AOA transcripts, making it by far the most active phylotype. It showed 94.7% similarity to Candidatus Nitrosomarinus catalina SPOT01, an isolated strain from the coast off California. While it can be expected that only a subset of a functional gene pool is expressed at a point in time (e.g. nitrogenase genes; Severin et al., 2015), the consistent expression in aOTU1 is

112

intriguing as it points to a high environmental relevance and tolerance by this phylotype. Drivers of community composition of ammonia

oxidizers Site, environment (river or estuary), and salinity correlated with the composition of both AOA and AOB communities indicating that these are strong drivers of the composition of ammonia oxidizers. Interestingly, PON and depth correlated with AOA composition, and depth correlated with the composition of amoA expressing AOA, reflecting that several aOTUs were found exclusively in deep water (before the storm). Similarly, AOA clades showed distinct depth distributions in the Southern California Bight (Beman et al., 2008). Hence, depth appears a strong driver of AOA composition. In contrast, the composition of AOB was correlated with temperature, possibly suggesting season as an important driver. These patterns may reflect differences in metabolism between AOA and AOB. For instance, Nitrosopomulis maritimus has a different ammonia oxidation pathway than that in bacteria rendering these archaea an advantage in low nutrient environments (Walker et al., 2010). However, at present the scarcity of data prevents trustworthy identifications of factors determining the prevalence of specific ammonia oxidizer taxa. Concluding remarks

Our study documents a prominent role of active ammonia oxidizing bacteria in the coastal Baltic, which is in contrast to oceanic or marine coastal waters, and particularly widespread amoA gene expression by an AOA phylotype related to Nitrosomarinus. Our study highlights coastal regions of the Baltic Sea as dynamic sites of extensive pelagic

nitrification, which may affect local food web dynamics and loss of N mediated by denitrification. This should be taken into account in future modelling of N cycling in the Baltic Sea.

Experimental procedures

Sampling

Samples were collected during a cruise to the Bay of Gdańsk and Vistula River (R/V Alkor, January 31 – February 13, 2015) and cruises to the Öre Estuary (Bothnian Sea, R/V Lotty April 20 – 24 and August 3 – 8, 2015; Fig. 1, Table 1). Samples for environmental variables, nitrification, and nucleotide extraction were obtained from surface and bottom waters and occasionally at intermediate depths using 5- or 10- L Niskin bottles. Moreover, at station VE02 (22 m) water was collected from 30 - 50 cm above intact sediment cores obtained with a Multi Corer (MUC).

Environmental variables

Water column measurements in the Bay of Gdánsk were performed with a Seabird 911plus CTD-system whereas a Seabird SBE19plus CTD-system was used in Öre Estuary. Concentrations of nitrite, nitrate, and ammonium were measured from filtered water samples (Öre estuary: Acrodisc 32 mm Syringe Filter with 0.2 μm Supor Membrane, sterile; Vistula estuary: Whatman GF/F, nominal pore size 0.7 µm) with a continuous segmented flow analyzer (QuAAtro, Seal Analytical) following Grasshoff et al. (1999) and HELCOM guidelines. For particulate organic nitrogen (PON) and carbon (POC) determinations, up to 2 L of water was filtered onto pre-combusted GF/F filters (Whatman, 3 h at 450°C), which were then stored frozen. In

113

the laboratory, the filters were dried at 60°C overnight, packed into tin capsules, and measured with an element analyser (Thermo Flash 2000). Calibration for PON and POC content was done with acetanilide (Merck; 10.36 % N; 71.09 % C).

Nitrification

Nitrification rates were determined using the 15N-NH4+ tracer method (Ward, 2005; Veuger et al., 2013; Damashek et al., 2016), as described in detail by Bartl et al. (submitted). Briefly, polycarbonate bottles were filled with water, avoiding air bubbles, and sealed with a butyl septum. The samples were then amended with 50 nmol L-1 15N-NH4Cl (99 atom% 15N) to yield an enrichment of < 30 % and ensure measurement of in situ nitrification rates. Three t0-bottles were filtered (200 mbar) immediately through precombusted GF/F filters (Whatman, 3 h at 450°C) while the remaining triplicates were incubated for 5 – 7 h in the dark at in situ temperature before filtration. Filtrates were stored frozen at - 20°C until analysis. The 15N content of nitrite plus nitrate was measured at the Leibniz-Institute for Baltic Sea Research Warnemünde via the denitrifier method (Sigman et al., 2001; Casciotti et al., 2002) using a Finnigan GasBench II and a Delta V advantage (Thermo). Nitrate isotope references (IAEA-N3 and USGS 34) were used for calibration and the accuracy of the isotope measurements was ± 0.14 ‰. Nitrification rates were calculated according to Veuger et al. (2013) and corrected for the percent labelling. Since both, the 15N content of nitrite and nitrate were measured the calculated nitrification rate is a bulk nitrification rate including both ammonium oxidation and nitrite oxidation.

Quantification of AOA and AOB genes and transcripts by digital droplet PCR (ddPCR)

In ddPCR, the sample and PCR assay mixture is divided into a very large number of separate small volume reactions (droplets), such that there is either zero or one target molecule present in any individual reaction. Thermal cycling is, thereafter, performed to endpoint. The distribution of target DNA molecules among the partitions follows Poisson statistics, and an absolute target sequence quantity can be estimated without the use of a standard (cf. Vogelstein and Kinzler 1999; Pinheiro et al., 2012; Dong et al., 2014). AmoA genes and transcripts from AOA and AOB (β-proteobacteria) were quantified from extracted DNA and RNA using a Bio-Rad ddPCR system and the primer sets amoA-1F(5’-GGGGTTTCTACTGGTGGT-3’) and amoA-2R (5’-CCCCTCKGSAAAGCCTTCTTC-3’) (Rotthauwe et al., 1997) and Arch-amoAF (5’-STAATGGTCTGGCTTAGACG-3’) and Arch-amoAR (5’-GCGGCCATCCATCTGTATGT-3’) (Francis et al., 2005), respectively. Correct length of PCR products were verified by agarose gel-electrophoresis prior to ddPCR. Samples were measured in triplicate PCR reactions. Each reaction mixture consisted of 10 µl Evagreen ddPCR mix (Bio-Rad), 200 nM of each primer, BSA (0.5 µg µl-1) and ~20 ng template. The mixture was loaded with 70 µl Evagreen droplet generation oil into the droplet generator. PCR was performed in a T100 thermal cycler using a profile of 95°C for 10 min, followed by 40 cycles of 94°C for 30 s and 60°C (AOA) or 53°C (AOB) for 60 s, 1 cycle of 98°C for 10 min, and ending at 4°C. Droplets were read on the droplet reader and data analyzed using the QuantaSoft software (Bio-Rad). Quantification was presented as the number of target molecules per μl of PCR mixture and converted to copy number per liter seawater using the volume of seawater

114

filtered and the total amount of DNA or RNA extracted from a sample, assuming equal extraction efficiency among samples. The amoA expression abundances were corrected using the reverse-transcription factor (see below).

PCR amplification and sequencing of ammonia oxidation (amoA) genes

Nucleic acids were extracted using the Allprep DNA/RNA mini kit (Qiagen). DNA was purified using the Genomic DNA purification and concentrator kit (Zymo) and RNA using the RNA purification and concentrator kit (Zymo) with build-in DNAse treatment. DNA and RNA were quantified using Quant-IT PicoGreen and Quant-IT RiboGreen (Invitrogen), respectively. PCRs on extracted RNA controlled for complete DNA digestion. cDNA was synthesized using the TaqMan Reverse Transcription (RT) Reagents (TaqMan) and amoA-specific reverse primers amoA2R and Arch-amoAR. The RT products were quantified using PicoGreen and the efficiency of each RT reaction (reverse-transcription factor) was calculated from the amount of template added to the RT divided by the product. AmoA was amplified from both DNA and cDNA using barcoded β-proteobacterial and archaeal primers (see above). Unfortunately, amoA genes were only amplifiable from a single cDNA sample using the AOB primers, despite numerous PCR optimization steps involving different concentrations of MgCl2, bovine serum albumin and thermal cycling conditions. We speculate that the different chemistry and the high sensitivity enabled successful amplification of AOB transcripts using ddPCR (Dong et al., 2014), and not by conventional PCR. Primers targeting the amoAC of γ-proteobacteria (Purkhold et al., 2000) were

tested but without successful amplification of samples. Each 25 µl PCR reaction contained 0.25 µl MyTaq Polymerase (Bioline Reagents Ltd), 625 nM of each primer and 9-46 ng template. DNA from Nitrosomonas europaea (DSM 28437, German Collection of Microorganisms and Cell Cultures) served as positive control for the bacterial amoA PCR. Triplicate PCR products per sample were pooled, purified (Agentcourt AMPure XP, Beckman Coulter), quantified, and the samples were then pooled in equimolar concentrations and paired-end sequenced on a 2x300 bp Illumina MiSeq platform (Eurofins Genomics, Ebersberg, Germany). Sequences are deposited in the EMBL database under accession number SAMN07774291.

Sequence analysis and statistics

Sequence reads were de-multiplexed, trimmed, filtered, and clustered using Qiime (Caporaso et al., 2010). The paired-end reads were joined using fastq-join (Aronesty, 2011). Only sequences with a phred-score of ≥20 were considered for downstream analysis. The sequences were clustered at 97% similarity using UCLUST (Edgar, 2010). Maximum likelihood trees were created in Mega 7.0 (Kumar et al., 2016) using Tamura-Nei models and 500 bootstrappings. Archaeal forward and reverse reads could not be joined due to the large amplicon length (635 bp) relative to the Illumina read length of 300 bp. Hence, due to the lack of overlap the forward and reverse reads were analyzed separately. Each stretch (stretch A and B) were clustered at 97% similarity and representative sequences analyzed in a phylogenetic tree (data not shown). Stretch B clustered with bacterial amoA gene sequences, suggesting that this region of the amoA gene was rather conserved. In contrast, stretch A sequences

115

showed high diversity and aligned with known archaeal amoA genes. Stretch A, with a length of 258 bp, was therefore chosen for the downstream analysis. Operational taxonomic unit (OTU) data were normalized using DEseq2 (Love et al., 2014). Principal component and Redundancy analyses were done using the R-package Vegan (R Core Team, 2015). Permutational ANOVA was used to test for differences between communities based on Bray-Curtis dissimilarities. Prior to the analysis, OTU tables were filtered to only include OTUs occurring more than 40 times in the entire dataset.

Correlations between environmental variables, AOA and AOB gene abundances and nitrification were tested using Pearson’s product-moment correlation. Environmental parameters were tested using the Mann-Whitney U-test.

Acknowledgements This research was supported by the BONUS Blueprint and BONUS Cocoa projects that have received funding from BONUS, the joint Baltic Sea research and development programme (Art 185), funded jointly from the European Union’s Seventh Programme for research, technological development and demonstration as well as from The Danish Council for Strategic Research (3051-00002B) and the German Ministry of Education and Science (03F0683A). We thank Mads Obi Bergsten for help with sequence analyses, Christin Bennke and Sarah Beier for guidance on DNA and RNA extraction, Morten Schiøtt for guidance in ddPCR, Iris Liskow, Christian Burmeister, and Henrik Larsson for help with nutrient analyses, and the anonymous reviewers for constructive suggestions that

significantly improved the manuscript. The authors declare no conflict of interest. References

Abell, G.C.J., Revill, A.T., Smith, C., Bissett, A.P., Volkman, J.K., and Robert, S.S. (2009) Archaeal ammonia oxidizers and nirS-type denitrifiers dominate sediment nitrifying and denitrifying populations in a subtropical macrotidal estuary. ISME J 4: 286-300.

Ahlgren, N.A., Chen, Y., Needham, D.M., Parada, A.E., Sachdeva, R., Trinh, V. et al. (2017) Genome and epigenome of a novel marine Thaumarchaeota strain suggest viral infection, phosphorothioation DNA modification and multiple restriction systems. Environmental Microbiology 19: 2434-2452.

Andersson, M.G.I., Rijswijk, P.v., and Middelburg, J.J. (2006) Uptake of dissolved inorganic nitrogen, urea and amino acids in the Scheldt estuary: comparison of organic carbon and nitrogen uptake. Aquatic Microbial Ecology 44: 303-315.

Aronesty, E. (2011) Command-line tools for processing biological sequencing data. ea-utils.

Bano, N., and Hollibaugh, J.T. (2000) Diversity and Distribution of DNA Sequences with Affinity to Ammonia-Oxidizing Bacteria of the β Subdivision of the Class Proteobacteria in the Arctic Ocean. Applied and Environmental Microbiology 66: 1960-1969.

Bayer, B., Vojvoda, J., Offre, P., Alves, R.J.E., Elisabeth, N.H., Garcia, J.A.L. et al. (2015) Physiological and genomic characterization of two novel marine thaumarchaeal strains indicates niche differentiation. ISME J 10: 1051–1063.

Beman, J.M., Popp, B.N., and Francis, C.A. (2008) Molecular and biogeochemical evidence for ammonia oxidation by marine Crenarchaeota in the Gulf of California. ISME J 2: 429-441.

Beman, J.M., Sachdeva, R., and Fuhrman, J.A. (2010) Population ecology of nitrifying archaea and bacteria in the Southern California Bight. Environmental Microbiology 12: 1282-1292.

Berg, C., Vandieken, V., Thamdrup, B., and Jurgens, K. (2015) Significance of archaeal nitrification in hypoxic waters of the Baltic Sea. ISME J 9: 1319-1332.

Bernes, C. (2005) Change Beneath the Surface. An In-Depth Look at Sweden's Marine

116

Environment. Fälth & Hässler, Värnamo, Sweden Monitor 19: 38-47.

Bianchi, M., Feliatra, and Lefevre, D. (1999) Regulation of nitrification in the land-ocean contact area of the Rhône River plume (NW Mediterranean). Aquatic Microbial Ecology 18: 301-312.

Bristow, L.A., Sarode, N., Cartee, J., Caro-Quintero, A., Thamdrup, B., and Stewart, F.J. (2015) Biogeochemical and metagenomic analysis of nitrite accumulation in the Gulf of Mexico hypoxic zone. Limnology and Oceanography 60: 1733-1750.

Caffrey, J.M., Bano, N., Kalanetra, K., and Hollibaugh, J.T. (2007) Ammonia oxidation and ammonia-oxidizing bacteria and archaea from estuaries with differing histories of hypoxia. ISME J 1: 660-662.

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K. et al. (2010) QIIME allows analysis of high-throughput community sequencing data. Nature Methods 7: 335-336.

Casciotti, K.L., Sigman, D.M., Hastings, M.G., Böhlke, J.K., and Hilkert, A. (2002) Measurement of the Oxygen Isotopic Composition of Nitrate in Seawater and Freshwater Using the Denitrifier Method. Analytical Chemistry 74: 4905-4912.

Chang, Y., Fan, J., Su, J., Ming, H., Zhao, W., Shi, Y. et al. (2017) Spatial Abundance, Diversity, and Activity of Ammonia-Oxidizing Bacteria in Coastal Sediments of the Liaohe Estuary. Current Microbiology 74: 632-640.

Clark, D.R., Rees, A.P., and Joint, I. (2008) Ammonium regeneration and nitrification rates in the oligotrophic Atlantic Ocean: Implications for new production estimates. Limnology and Oceanography 53: 52-62.

Damashek, J., Casciotti, K.L., and Francis, C.A. (2016) Variable Nitrification Rates Across Environmental Gradients in Turbid, Nutrient-Rich Estuary Waters of San Francisco Bay. Estuaries and Coasts 39: 1050-1071.

Dell'Anno, A., Fabiano, M., Duineveld, G.C.A., Kok, A., and Danovaro, R. (1998) Nucleic Acid (DNA, RNA) Quantification and RNA/DNA Ratio Determination in Marine Sediments: Comparison of Spectrophotometric, Fluorometric, and High-Performance Liquid Chromatography Methods and Estimation of Detrital DNA. Applied and Environmental Microbiology 64: 3238–3245.

Dong, L., Meng, Y, Wang, J., and Liu, Y. (2014) Evaluation of droplet digital PCR for characterizing plasmid reference material used for quantifying ammonia oxidizers and denitrifiers. Analytical and Bioanalytical Chemistry 406: 1701-1712

Edgar, R.C. (2010) Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26: 2460-2461.

Enoksson, V. (1986) Nitrification Rates in the Baltic Sea - Comparison of 3 Isotope Techniques. Applied and Environmental Microbiology 51: 244-250.

Fan, H., Bolhuis, H., and Stal, L. (2015) Nitrification and nitrifying bacteria in a coastal microbial mat. Frontiers in Microbiology 6:1367.

Francis, C.A., O'Mullan, G.D., and Ward, B.B. (2003) Diversity of ammonia monooxygenase (amoA) genes across environmental gradients in Chesapeake Bay sediments. Geobiology 1: 129-140.

Francis, C.A., Roberts, K.J., Beman, J.M., Santoro, A.E., and Oakley, B.B. (2005) Ubiquity and diversity of ammonia-oxidizing archaea in water columns and sediments of the ocean. Proceedings of the National Academy of Sciences of the United States of America 102: 14683-14688.

Gattuso, J.-P., Frankignoulle, M., and Wollast, R. (1998) Carbon and carbonate metabolism in coastal aquatic ecosystems. Annual Review of Ecology and Systematics 29: 405-434.

Glibert, P.M., Wilkerson, F.P., Dugdale, R.C., Raven, J.A., Dupont, C.L., Leavitt, P.R. et al. (2016) Pluses and minuses of ammonium and nitrate uptake and assimilation by phytoplankton and implications for productivity and community composition, with emphasis on nitrogen-enriched conditions. Limnology and Oceanography 61: 165-197.

Goldberg, S.J., Nelson, C.E., Viviani, D.A., Shulse, C.N., and Church, M.J. (2017) Cascading influence of inorganic nitrogen sources on DOM production, composition, lability and microbial community structure in the open ocean. Environmental Microbiology . Published online, doi: 10.1111/1462-2920.13825.

Grasshoff, K., Kremling, K., and Ehrhardt, M. (1999) Methods of Seawater Analysis: VCH Publishers, pp. 159-228

Herlemann, D.P.R., Lundin, D., Andersson, A.F., Labrenz, M., and Jürgens, K. (2016) Phylogenetic Signals of Salinity and Season in Bacterial Community Composition Across the

117

Salinity Gradient of the Baltic Sea. Frontiers in Microbiology 7: 1883.

Hietanen, S., Jäntti, H., Buizert, C., Jürgens, K., Labrenz, M., Voss, M., and Kuparinen, J. (2012) Hypoxia and nitrogen processing in the Baltic Sea water column. Limnology and Oceanography 57: 325-337.

Hollibaugh, J.T., Gifford, S., Sharma, S., Bano, N., and Moran, M.A. (2011) Metatranscriptomic analysis of ammonia-oxidizing organisms in an estuarine bacterioplankton assemblage. ISME J 5: 866-878.

Horak, R.E.A., Qin, W., Schauer, A.J., Armbrust, E.V., Ingalls, A.E., Moffett, J.W. et al. (2013) Ammonia oxidation kinetics and temperature sensitivity of a natural marine community dominated by Archaea. ISME J 7: 2023-2033.

Karl, D. M., Knauer, G. A., Martin, J. H. and Ward, B. B. (1984) "Bacterial chemolithotrophy in the ocean is associated with sinking particles." Nature 309: 54.

Kim, O.S., Junier, P., Imhoff, J.F., and Witzel, K.P. (2008) Comparative analysis of ammonia monooxygenase (amoA) genes in the water column and sediment-water interface of two lakes and the Baltic Sea. Fems Microbiology Ecology 66: 367-378.

Konneke, M., Bernhard, A.E., de la Torre, J.R., Walker, C.B., Waterbury, J.B., and Stahl, D.A. (2005) Isolation of an autotrophic ammonia-oxidizing marine archaeon. Nature 437: 543-546.

Kumar, S., Stecher, G., and Tamura, K. (2016) MEGA7: Molecular Evolutionary Genetics Analysis Version 7.0 for Bigger Datasets. Molecular Biology and Evolution 33: 1870-1874.

Labrenz, M., Sintes, E., Toetzke, F., Zumsteg, A., Herndl, G.J., Seidler, M., and Jurgens, K. (2010) Relevance of a crenarchaeotal subcluster related to Candidatus Nitrosopumilus maritimus to ammonia oxidation in the suboxic zone of the central Baltic Sea. ISME J 4: 1496-1508.

Lee, H.-G.1, Kim, H.M., Min, J., Kim, K., Park, M. G., Jeong. H.J., and Kim, K.Y. (2017) An advanced tool, droplet digital PCR (ddPCR), for absolute quantification of the red-tide dinoflagellate, Cochlodinium polykrikoides Margalef (Dinophyceae). Algae 32: 189-197.

Llorens-Mares, T., Yooseph, S., Goll, J., Hoffman, J., Vila-Costa, M., Borrego, C.M. et al. (2015) Connecting biodiversity and potential functional role in modern euxinic environments by microbial metagenomics. ISME J 9: 1648-1661.

Love, M.I., Huber, W., and Anders, S. (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology 15: 550.

Martens-Habbena, W., Berube, P.M., Urakawa, H., de la Torre, J.R., and Stahl, D.A. (2009) Ammonia oxidation kinetics determine niche separation of nitrifying Archaea and Bacteria. Nature 461: 976-979.

Mincer, T.J., Church, M.J., Taylor, L.T., Preston, C., Karl, D.M., and DeLong, E.F. (2007) Quantitative distribution of presumptive archaeal and bacterial nitrifiers in Monterey Bay and the North Pacific Subtropical Gyre. Environmental Microbiology 9: 1162-1175.

Newell, S.E., Fawcett, S.E., and Ward, B.B. (2013) Depth distribution of ammonia oxidation rates and ammonia oxidizer community composition in the Sargasso Sea. Limnology and Oceanography 58: 1491-1500.

Nygaard, S., Hu, H., Li, C., Schiøtt, M., Chen, Z., Yang, Z., et al. (2016). Reciprocal genomic evolution in the ant–fungus agricultural symbiosis. Nature Communications 7: 12233.

O'Mullan, G.D., and Ward, B.B. (2005) Relationship of Temporal and Spatial Variabilities of Ammonia-Oxidizing Bacteria to Nitrification Rates in Monterey Bay, California. Applied and Environmental Microbiology 71: 697-705.

Pinheiro, L.B., Coleman, V.A., Hindson, C.M., Herrmann, J., Hindson, B.J., Bhat, S., and Emslie, K.R. (2012) Evaluation of a droplet digital polymerase chain reaction format for DNA copy number quantification. Analytical Chemistry 84: 1003-1011

Ploug, H., and Bergkvist, J. (2015) Oxygen diffusion limitation and ammonium production within sinking diatom aggregates under hypoxic and anoxic conditions. Marine Chemistry 176: 142-149.

Purkhold, U., Pommerening-Roser, A., Juretschko, S., Schmid, M.C., Koops, H.P., and Wagner, M. (2000) Phylogeny of All Recognized Species of Ammonia Oxidizers Based on Comparative 16S rRNA and amoA Sequence Analysis: Implications for Molecular Diversity Surveys. Applied and Environmental Microbiology 66: 5368-5382.

R Core Team (2015) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Rotthauwe, J.H., Witzel, K.P., and Liesack, W. (1997) The ammonia monooxygenase structural

118

gene amoA as a functional marker: Molecular fine-scale analysis of natural ammonia-oxidizing populations. Applied and Environmental Microbiology 63: 4704-4712.

Rönnberg, C., and Bonsdorff, E. (2004) Baltic Sea eutrophication: area-specific ecological consequences. Hydrobiologia 514: 227-241.

Sandberg, J., Andersson, A., Johansson, S., and Wikner, J. (2004) Pelagic food web structure and carbon budget in the northern Baltic Sea: potential importance of terrigenous carbon. Marine Ecology Progress Series 268: 13-29.

Schleper, C., Jurgens, G., and Jonuscheit, M. (2005) Genomic studies of uncultivated archaea. Nature Reviews Microbiology 3: 479-488.

Severin, I., Bentzon-Tilia, M., Moisander, P.H., and Riemann, L. (2015) Nitrogenase expression in estuarine bacterioplankton influenced by organic carbon and availability of oxygen. FEMS Microbiology Letters 362: fnv105.

Shiozaki, T., Ijichi, M., Isobe, K., Hashihama, F., Nakamura, K.-i., Ehama, M. et al. (2016) Nitrification and its influence on biogeochemical cycles from the equatorial Pacific to the Arctic Ocean. ISME J 10: 2184–2197.

Sigman, D.M., Casciotti, K.L., Andreani, M., Barford, C., Galanter, M., and Böhlke, J.K. (2001) A Bacterial Method for the Nitrogen Isotopic Analysis of Nitrate in Seawater and Freshwater. Analytical Chemistry 73: 4145-4153.

Silvennoinen, H., Hietanen, S., Liikanen, A., Stange, C.F., Russow, R., Kuparinen, J., and Martikainen, P.J. (2007) Denitrification in the river estuaries of the northern Baltic Sea. Ambio 36: 134-140.

Simon, M., Grossart, H.P., Schweitzer, B., and Ploug, H. (2002) Microbial ecology of organic aggregates in aquatic ecosystems. Aquatic Microbial Ecology 28: 175-211.

Smith, J.M., Casciotti, K.L., Chavez, F.P., and Francis, C.A. (2014) Differential contributions of archaeal ammonia oxidizer ecotypes to nitrification in coastal surface waters. ISME J 8: 1704-1714.

Twomey, L.J., Piehler, M.F., Paerl, H.W. (2005) Phytoplankton uptake of ammonium, nitrate and urea in the Neuse River Estuary, NC, USA. Hydrobiologia 533: 123-134.

Utaker, J.B., Bakken, L., Jiang, Q.Q., and Nes, I.F. (1996) Phylogenetic analysis of seven new isolates of ammonia-oxidizing bacteria based

on 16S rRNA gene sequences. Systematic and Applied Microbiology 18: 549-559.

Venter, J.C., Remington, K., Heidelberg, J.F., Halpern, A.L., Rusch, D., Eisen, J.A. et al. (2004) Environmental Genome Shotgun Sequencing of the Sargasso Sea. Science 304: 66-74.

Vetterli, A., Hietanen, S., and Leskinen, E. (2016) Spatial and temporal dynamics of ammonia oxidizers in the sediments of the Gulf of Finland, Baltic Sea. Marine Environmental Research 113: 153-163.

Veuger, B., Pitcher, A., Schouten, S., Sinninghe Damsté, J.S., and Middelburg, J.J. (2013) Nitrification and growth of autotrophic nitrifying bacteria and Thaumarchaeota in the coastal North Sea. Biogeosciences 10: 1775-1785.

Vogelstein, B., and Kinzler, K.W. (1999) Digital PCR. Proceedings of the National Academy of Sciences of the United States of America 96: 9236-9241.

Voytek, M.A., and Ward, B.B. (1995) Detection of Ammonium-Oxidizing Bacteria of the Beta-Subclass of the Class Proteobacteria in Aquatic Samples with the Pcr. Applied and Environmental Microbiology 61: 2811-2811.

Walker, C.B., de la Torre, J.R., Klotz, M.G., Urakawa, H., Pinel, N., Arp, D.J. et al. (2010) Nitrosopumilus maritimus genome reveals unique mechanisms for nitrification and autotrophy in globally distributed marine crenarchaea. Proceedings of the National Academy of Sciences of the United States of America 107: 8818-8823.

Wang, Y., Naumann, U., Wright, S.T., and Warton, D.I. (2012) mvabund– an R package for model-based analysis of multivariate abundance data. Methods in Ecology and Evolution 3: 471-474.

Wankel, S.D., Mosier, A.C., Hansel, C.M., Paytan, A., and Francis, C.A. (2011) Spatial variability in nitrification rates and ammonia-oxidizing microbial communities in the agriculturally impacted Elkhorn Slough estuary, California. Applied and Environmental Microbiology 77: 269-280.

Ward, B.B. (1990) Kinetics of ammonia oxidation by a marine nitrifying bacterium: Methane as a substrate analogue. Microbial Ecology 19: 211-225.

Ward, B.B. (1996) Nitrification and denitrification: Probing the nitrogen cycle in aquatic environments. Microbial Ecology 32: 247-261.

Ward, B.B. (2005) Temporal variability in nitrification rates and related biogeochemical

119

factors in Monterey Bay, California, USA. Marine Ecology Progress Series 292: 97-109.

Ward, B.B. (2013) Nitrification. In Reference Module in Earth Systems and Environmental Sciences, Elsevier, pp. 1-8.

Ward, B.B., and Kilpatrick, K.A. (1990) Relationship between substrate concentration and oxidation of ammonium and methane in a stratified water column. Continental Shelf Research 10: 1193-1208.

Wuchter, C., Abbas, B., Coolen, M.J.L., Herfort, L., van Bleijswijk, J., Timmers, P. et al. (2006) Archaeal nitrification in the ocean. Proceedings

of the National Academy of Sciences of the United States of America 103: 12317-12322.

Zhang, Y., Xie, X., Jiao, N., Hsiao, S.S.Y., and Kao, S.J. (2014) Diversity and distribution of amoA-type nitrifying and nirS-type denitrifying microbial communities in the Yangtze River estuary. Biogeosciences 11: 2131-2145.

Zheng, Y.L., Hou, L.J., Newell, S., Liu, M., Zhou, J.L., Zhao, H. et al. (2014) Community Dynamics and Activity of Ammonia-Oxidizing Prokaryotes in Intertidal Sediments of the Yangtze Estuary. Applied and Environmental Microbiology 80: 408-4

120

Supporting Information

121

122

Supporting information

Extensive nitrification and active ammonia oxidizers in two contrasting coastal systems of the Baltic Sea

Elisabeth Happel1, Ines Bartl2, Maren Voss2, Lasse Riemann1*

*Corresponding author: Lasse Riemann (lriemann@bio.ku.dk), Strandpromenaden 5, 3000 Helsingør (DK), phone +4535321959.

Figure S1. Maximum likelihood tree of archaeal partial amoA nucleotide sequences (288 bp). OTUs clustered at 97% sequence similarities shown with relevant closest-relative sequences from NCBI. Bootstrap values are based on 500 replications.

123

Figure S2. Maximum likelihood of β-proteobacterial partial amoA nucleotide sequences (288 bp). OTUs clustered at 97% sequence similarities shown with relevant closest-relative sequences from NCBI. Bootstrap values are based on 500 replications.

124

Figure S3. Correlations between environmental data, amoA gene and transcript abundances and nitrification in the Bay of Gdańsk (A) and Öre Estuary (B). Significant (p <0.05) correlations investigated through Pearson’s product-moment correlation shown. Colors indicate Pearson’s correlation coefficient, positive correlations (blue) and negative correlations (red).

125

Figure S4. Redundancy analysis (RDA) to identify potential linkages between ammonia oxidizing archaeal (top) and bacterial (bottom) community composition and environmental variables. Red and empty circles indicate operational taxonomic units and samples, respectively. Note that two samples were excluded from the analysis due to missing data points. Temp = temperature; Sali = salinity; ammo = ammonium; PON = particulate organic nitrogen.

126

127

2

The effects of riverine DOM on microbial composition

and function

List of papers included in this thesis

I. T. Markussen, E. M. Happel, J. E. Teikari, V. Huchaiah, J. Alneberg, A. Andersson, K. Sivonen, L. Riemann, M. Middelboe and V. Kisand (2018) Coupling biogeochemical process rates and metagenomic blueprints of coastal bacterial assemblages in the context of environmental change. Submitted to Environmental Microbiology

II. E. M. Happel, T. Markussen, J. Teikari, V. Huchaiah, J. Alneberg, A. Andersson, K. Sivonen, M. Middelboe, V. Kisand and L. Riemann (2018) Effects of allocthonous DOM input on microbial composition and nitrogen cycling genes at two contrasting estuarine sites. Submitted to Environmental Microbiology Reports

III. E. Happel, I. Bartl, M. Voss and L. Riemann (2018) Extensive nitrification and active ammonia oxidizers in two contrasting coastal systems of the Baltic Sea. Environmental Microbiology, in press